Chapter 2 of Understanding Chemical Bonding: The development of A level students' understanding of the concept of chemical bonding
Constructivism in science education
§2.0: The alternative conceptions movement in science education
"there is an extensive literature that indicates that children come to their science classes with prior conceptions that may differ substantially from the ideas to be taught"
Driver, 1989, p.481
The present research has been undertaken within a tradition of work in science education which has been called the alternative conceptions movement, (A.C.M.), or constructivism. This tradition has been recognised as a specific field of research, so that Gilbert and Swift suggest (1985, p.682) that the A.C.M. has the characteristics of a Lakatosian research programme (c.f. Lakatos, 1970), and Solomon (1994, p.7) refers to constructivism as a Kuhnian paradigm (c.f. Kuhn, 1970 {1962} ). In a similar vein, Matthews refers to the influence of constructivism "as if a period of Kuhnian normal science has descended upon the science and mathematics education communities" (1993, p.363).
The term 'constructivism' has also been adopted by some to describe the much broader movement of workers undertaking qualitative, interpretative research (§4.1), as in contrast to what Beld has called "the conventional positivist paradigm in social science research" (1994, p.99, c.f. §4.1). However, when the term 'constructivism' is used in this present work it will refer to the tradition of research in science education, rather than the broader meaning. This distinction is significant as constructivism has been criticised (Matthews, 1994; Suchting, 1992), due to the strong relativistic stance suggested by some thinkers – such as von Glaserfeld's 'radical constructivism' (e.g. 1989) – which implies that learners' alternative constructions of science have as much validity as those of scientists. Although this is an interesting theoretical position, it is not one shared by most of the constructivists working is science education (although see §2.7 below), where an ethnographic approach may be used to elicit learners' notions, but where the usual aim (sometimes, but not always, made explicit) is to inform the teaching of accepted science as mediated by the science curriculum.
At the time of writing this thesis it is generally accepted that – as the motto above reports – children and students come to science classes with alternative conceptions about many aspects of science, and the constructivist view is that these 'interfere' with the intended learning of curricular material (or provide substantive learning impediments in my term from chapter 1, §1.5.3). These beliefs are the basis for the "period of Kuhnian normal science" that Matthews refers to (1993, p.363). The possibility that the student has a 'blank mind' – the tabula rasa of a Baconian philosophy of science – is not considered tenable by constructivists (although research shows teachers' metaphors for the learning process are often quite compatible with it, {Fox, 1983; Tobin, 1990, pp.33-35} , and that almost half of student teachers surveyed thought that learners passively accumulate knowledge that is transmitted to them {Hennessy, 1993, p.8, and see also Linder, 1992} ).
There is a vast literature on learners' understanding of science (Gilbert, 1994), as is shown by the review of research into children's ideas by Driver et al. (1994a). The bibliography 'Research on Students' Conceptions in Science' compiled by Carmichael et al. (1990) cites well over a thousand papers. Pfundt and Duit's bibliography includes 132 studies referring to chemistry topics, 208 in biology, and 740 relating to topics that are largely within physics (Duit, 1991, p.71).
It is only possible to refer directly to a small part of this body of work, and in this review I consider the literature which is most relevant to the present study. In this chapter I discuss the literature referring to the general theoretical framework which has been used to illuminate and define the field of enquiry. In the subsequent chapter I turn to what was already known about learners' ideas of chemical bonding prior to the research reported herein, and I consider several other topics where I argue that the learners' ideas have consequences for understanding the bonding topic.
§2.1: Cognitive structure, progression and learning
In the previous chapter certain assumptions were made which are taken as axiomatic in the present study (§1.4). In summary: the existence of cognitive structure, its influence on the learner's behaviour (particularly verbal behaviour), and the ability to both construct and evaluate models of aspects of an individual's cognitive structure are all taken as reasonable foundations for the research.
These assumptions refer to something static – an individual's cognitive structure at one moment in time. The present study investigates developing understanding (§1.8), and is therefore concerned with changes in cognitive structure, and how they come about. In chapter 1 the concept of progression was considered, and was related to the nature of chemistry as a subject (§1.7). As this study concerns students who have been successful at G.C.S.E. level and are aiming to successfully tackle an A level chemistry examination, progression may be understood in terms of the expected appreciation of the bonding concept at these two examination levels, as in the analysis presented in appendix 4. It is in terms of such an analysis that learning may be judged as 'appropriate' or otherwise. Progression therefore involves such 'appropriate' learning, which implies changes in cognitive structure: that is changes in the facts, concepts, propositions, theories, and raw perceptual data that the learner has available to her at any point in time, and the manner in which it is arranged (§1.4.1).
§2.1.1: The learning process and learning impediments
Chapter 1 also considered types of learning impediments: in other words reasons why a motivated learner placed in a teaching situation might not undertake appropriate learning. In terms of cognitive structures, this would mean that after instruction the learner's cognitive structure did not sufficiently match that intended. The simple typology of learning impediments presented may be seen to rest upon a major assumption: that is that a key variable in the learning process is the individual's cognitive structure before teaching occurs (e.g. Driver and Bell, 1986). In other words two learners faced with exactly the same lesson may learn differently from it, both quantitatively and qualitatively, depending on their existing arrangement of knowledge. This is one of the main assumptions of the constructivist position.
§2.1.2: Construction of knowledge
Constructivism is a perspective on learning, which views learning as an active process (§2.1.2). That is, the learner is seen to be pro-active rather than being the passive recipient of given knowledge (e.g. Driver and Bell, 1986, p.448, Pope and Gilbert, 1983, p.194). People have a predilection to make sense of their environment, and to arrange their memories of perceptions in some sort of pattern that acts as a framework for making sense of future experience (e.g. Osborne and Wittrock, 1983, p.492). It is also assumed that the construction process takes place through a series of steps of limited size. Although the 'construction of knowledge' is meant literally, the metaphor with constructing buildings [sic] is often seen as useful. As with building a house, building knowledge may be said to require good foundations, appropriate scaffolding and care to put each brick firmly in the correct place in the structure.
§2.2: Historical influences on the constructivist movement in science education
The origins of the constructivist perspective are ancient, and can be traced back as least as far as Plato and Socrates; through the use of parables by Jesus; to Rousseau's emphasis on (a) learning from actual experience rather than through verbal instruction and (b) the importance of the child's existing stage of development; and to Dewey's perception of the 'organic' nature of knowledge (Clark, 1968, p.181, p. 187-8; Egan, 1984, pp.28; Evetts, 1973, p.33; Russell, 1961 (1946), p.775). Dewey's view was that education should be learner-centred and that knowledge was something people constructed (Evetts, 1973, p.33; Hyland, 1993, p.94).
Among more modern thinkers constructivism in science education has been influenced by the work of psychologists such as Piaget, Kelly, Vygotsky, Bruner and Ausubel.
§2.2.1: Piaget
Jean Piaget is best known for proposing his 'stage-theory' of development, which has been widely reported (for example, Beard, 1969; Crain, 1992). Brown summarises this 'ages and stages' approach as "children of a given age are more likely to demonstrate similarity of [mental] structures that children of different ages" (1977, p.82), although he points out that this is just one aspect of Piaget's overall theory of genetic epistemology (i.e. the development of knowledge). S tage theory has been immensely influential, for instance as a starting point for workers attempting to match the science curriculum against what can reasonably be expected of learners at different ages (e.g. Shayer and Adey, 1981). So, the suggestion that only about 50% of people ever achieve Piaget's highest stage of formal operations (Arlin, 1975, p.605) would have serious implications for what can be effectively taught in science classes. The Piagetian stages were intended to operate across all areas of knowledge regardless of context, and this has been heavily criticised (for example see the discussion in Bliss, 1993). Piaget himself recognised that although the stages might in part reflect a maturation of the nervous system, this would only provide a potential which required experience of the physical environment and social interaction ("the educational influences of a favourable social environment") for its realisation (Inhelder and Piaget, 1964, p.5). The present study is not framed within this stage-theory tradition, and I will not elaborate on this argument here. However, one aspect of the debate is relevant.
Some workers have suggested that there should be a 'fifth stage' beyond formal operations, which might be considered to relate to mature adult thought (Arlin, 1975; Kramer, 1983). If this were truly a stage in the Piagetian sense it could only be attained by those that had successfully passed through the stage of formal operations. This would be significant because the proposed characteristics of such post-formal thinking are that: (a) knowledge has a relativistic, non-absolute nature; (b) contradiction is accepted as part of reality; and (c) the integrative approach to thinking is a central feature (Castro and Fernández, 1987, p.443). Kincheloe refers to "modes of thinking which transcend the formal operational ability to formulate abstract conclusions, understand cause-effect relationships, and employ the traditional scientific method to explain reality" (1991, p.44). Riegel has argued that tolerance of ambiguity reflects a more developed cognitive stage, so that dialectical operations is a stage beyond formal operations (Buck-Morss, 1980, p.130). It would seem that this is just the kind of thinking that may be needed to cope with a subject such as chemistry where the learner must learn to accept a range of partial, complementary – and sometimes inconsistent – models (§1.3.1, §1.7.1). If the sequential stage-theory approach were to be accepted, then we might expect that many A level students would not cope with such a subject as chemistry. However, those who reject the Piagetian stages would consider that there is no reason to believe that most learners can not develop 'post-formal thinking'.
The two aspects of Piaget's work which are particularly relevant are his methodology and the nature of his data. Piaget developed an approach of using clinical interviews such that "each child's thought patterns are traced by a series of questions, each being dependent upon the the previous response given by the child" (Brown, 1977, p.89). Such an approach was later taken up within the alternative conceptions movement (§4.6). Piaget published extensively, including details of some of the speech utterances of his young subjects. These data illustrate that the young child's thinking about the world may seem illogical, irrational and even contradictory to adults. In particular, Piaget highlighted the animistic and anthropomorphic nature of much of the reasoning of young children (discussed further in chapter 3, §3.1.4, and appendix 7, §A7.6). Further, they illustrate that children who have not undertaken formal instruction may still have constructed their own ideas about phenomena they experience in the world, and their own meanings for words as they acquire language (e.g. Piaget, 1973 {1929} ). Piaget wrote of the 'myth' of the sensory (or even perceptual) origins of scientific knowledge, and emphasised that the role of intelligence was to 'transform'. He believed that knowledge is formed by operating on perceptions with logico-mathematical frameworks (Piaget, 1972 {1970} ), that is aspects of existing cognitive structure (§1.4.1). These are key concerns for the science education community, and a focus for research within the constructivist tradition (c.f. §2.1.3).
Pope and Gilbert (1983) have commented on the constructivist assumptions inherent in much of Piaget's work (p.195.) and have described the "essence" of Piaget's epistemology as being "constructivist and relativist" (p.196). The active role of the learner as constructors of knowledge in Piaget's work has also been recognised by Driver and her coworkers (e.g. Driver and Easley, 1978; Driver et al., 1994c, p.3).
§2.2.2: Vygotsky
Piaget, then, assumed that the learner was an active constructor of knowledge, and his perspective focuses on the learner's actions on his or her environment. Vygotsky was contemporaneous to Piaget, but considered social interaction to be a central part of learning. Whereas Piaget's research programme was one of genetic epistemology (finding the cognitive-development sequence that each individual person would be expected to pass through), and took the view that "development explains learning" (Piaget, 1964, p.176), Vygotsky's programme was sociohistorical, that is it took the perspective that human psychological developments are mediated by culture and contingent on history (Cole, 1990, p.91). Vygotsky believed that from the age of about two years development is closely influenced by the young learner's interactions with other minds (Crain, 1992, p.199, p.211).
Vygotsky focuses on word-meaning as a useful unit for analysis (believing this to be the "unit of verbal thought that is further unanalysable and yet retains the properties of the whole", Vygotsky, 1986 {1934} , p.5). He pointed out that a word represents a generalisation (p.6). Fodor (1972) has suggested that Vygotsky's central thesis was that word meanings evolve as the child develops (p.86). For Vygotsky words were tools of thought (Vygotsky, 1986 {1934} , p.107), and were the essential tools for higher level thinking (p.251, Newman and Holzman, 1993, p.132), so that "real" concepts were not possible without them (Vygotsky, 1986 {1934} , p.107).
For Vygotsky, language was the medium in which teaching takes place, and from which the learner constructs a way of thinking (Edwards and Mercer, 1987, p.20). Vygotsky considered thought and speech to have different origins, but thought that the acquisition of words through speech provided the tools for conceptual thinking to develop (Vygotsky, 1986 {1934} , p.83).
In the present study it is assumed that the concepts of interest (such as electronegativity) can be represented by words (such as 'electronegativity'). Vygotsky points out that word meaning is tied to the context of use (p.245). The main data collection technique used in this study is the interview: so that colearners are asked to formulate their thinking using words, and those words are recorded, transcribed and analysed. Vygotsky's focus on words and word-meanings will inform the type of transcription considered appropriate for the present research (§5.2.1).
Vygotsky's analysis was undertaken from a perspective which recognised the social context in which learning and development take place (Edwards and Mercer, 1987, p.19). He recognised word meaning as the unit of the social interchange that was itself central to development of higher cognitive functions (Vygotsky, 1986 (1934), pp.7-9; 1978, p.57, p.88.) Like Piaget (§2.2.1), Vygotsky saw 'internalisation' as a process whereby an originally external operation becomes "reconstructed" within the mind of the individual (Vygotsky, 1978, p.56), and he emphasised that teachers could could lead pupils to higher levels of conceptual understanding than they would otherwise achieve (Edwards and Mercer, 1987, p.20).
Perhaps because of his focus on the social context of learning Vygotsky was interested in how learners solved problems when assisted by adults, compared with their competence when working alone. Independent working – "without the assistance of others, without demonstrations, and without leading questions" (Vygotsky 1978, p.88) is – from an educational perspective – a contrived situation, and Vygotsky put emphasis on the extent to which a learner could extend beyond their unassisted performance when provided with suitable cues. Bruner (see below, §2.2.3) has developed this perspective and considers that the teacher acts "as a vicarious form of consciousness" whilst the student is mastering a new skill (Hennessy's, 1993, p.13).
Vygotsky introduced the term 'the zone of proximal development' (or Z.P.D.) to describe the sphere of activity where a learner could solve problems with the teacher's guidance, or in collaboration with peers, but not independently (1978, p.86; 1986 {1934} , p.187). He thought the Z.P.D. was a better indicator of intellectual potential than mental age itself (1986 {1934} , p.187). It is within the Z.P.D. that control of cognitive functions is transferred from the interpersonal plane to become truly intrapersonal (Newman and Holzman, 1993, pp.66-67), so that what the learner can achieve within the Z.P.D. with assistance, is what he or she will next be able to achieve unaided (Crain, 1992, p.215).
In the present research much of the data derives directly from such a context of teacher-colearner dialogue, and it is accepted that through the research process the colearners may indeed be led to new levels of conceptual understanding. For an example, appendix 8 presents an extract of interview transcript where colearner Noor constructed a case for expecting interactions between neutral molecules (i.e. what are known to chemists as van der Waals' forces). However rather than see this as an unfortunate intervention that perturbs the subject of the study, it is seen from a Vygotskyan perspective as the natural context of concept development. Some methodological approaches would suggest that the interviewer should not provide cues when attempting to find out how much the learner understands. Yet a Vygotskyan perspective might suggest that an approach which provided a scaffold within the learner's Z.P.D. might ultimately be more revealing. This point is considered more at the end of chapter 4 (§4.10.4).
Of particular importance for constructivism is the distinction Vygotsky makes between two classes of concepts that he terms spontaneous, those which "emerge from the child's own reflections on everyday experience", and those he labels scientific and which the learner meets through formal instruction (Kozulin, 1985, pp.xxxiii-xxxiv) – whether in science or any other part of the curriculum (Newman and Holzman, 1993, p.61). These latter concepts tend to be given verbal definitions and are taught explicitly, unlike spontaneous concepts (Newman and Holzman, 1993, p.61). Although spontaneous concepts do not necessarily remain tacit, the distinction in origin is very significant for Vygotsky,
"The child becomes conscious of his spontaneous concepts relatively late; the ability to define them in words, to operate with them at will, appears long after he has acquired the concepts. He has the concept (i.e., knows the object to which the concept refers), but is not conscious of his own act of thought. The development of a scientific concept, on the other hand, usually begins with its verbal definition and its use in nonspontaneous operations – with working on the concept itself. It starts life in the child's mind at the level that his spontaneous concepts reach only later."
Vygotsky 1986 {1934} , p.192
Vygotsky believed that conceptual development involved a process of convergence as the concrete becomes abstracted, and the abstract is made concrete (1986 {1934} , p. 193). Over time spontaneous concepts would acquire a formal structure and be open to conscious use, and formal scientific concepts would evolve connections with real experience (1986 {1934} , p.194) – indeed scientific concepts provide the frameworks within which a learner could become aware of his tacit spontaneous concepts (Crain, 1992, p.213). Vygotsky then presupposed the presence of cognitive structure,
"Concepts do not lie in the child's mind like peas in a bag, without any bonds between them. If that were the case, no intellectual operation requiring coordination of thoughts would be possible, nor would any general conception of the world. Not even separate concepts as such could exist; their very nature presupposes a system."
Vygotsky 1986 {1934} , p.197
For scientific concepts the 'structures' made up a system (Vygotsky 1986 {1934} , p. 205), and the meaning of the concept depended on its relationship to other concepts in the system (c.f. §2.10.2). One of the techniques used to collect data in the present research was concept mapping, which attempts to make explicit just this aspect of concepts (§4.9.1).
§2.2.3: Bruner
Vygotsky's work has been disseminated and developed by Bruner (e.g. Crain, 1992, p.220), who concludes from studies of language acquisition in infants that in general people tend to have similar forms of mental organisation (1987, p.87, c.f. §1.5.4).
Bruner worked from Vygotsky's (1986 {1934} ) observation that communication between minds had to be indirect, and took place through language; and in particular that thoughts had to be translated into words (p.252). Bruner (1987) sees our shared use of language as a means of ensuring that we are sharing meaning, that is of "constant transactional calibration" (p.87) so that we can understand one another's minds and one another's worlds (p.88). Usually in conversation we understand what the other is saying, and – if not – we are usually aware of this, and have accepted ways of checking on meanings – i.e. what he has described as ways of "calling for repairs in one another's utterances to assure such calibration" (p.87).
This is an assumption that is built into the present research: that it is possible to understand my colearners' ideas about chemistry by interviewing them. This may seem an obvious point, and perhaps even trivial. However, Kuhn has referred to different paradigms as being incommensurate: in other words providing no basis for comparison. This position has been taken to mean that scientists working within two different scientific paradigms can not discuss the relative merits of their position in any objective way (Phillips, 1987, p.22). This would be very serious for the present work as it would mean that a teacher could not persuade students holding alternative frameworks about science that their views were in error as there would be no common basis for a rational discussion. Popper (1994) has referred to this as the myth of the !amework, that is that "a rational and fruitful discussion is impossible unless the participants share a common framework of basic assumptions or, at least, unless they have agreed on such a framework for the purpose of the discussion" (pp.34-5). Popper criticises this 'myth' (1970, p.56), although he accepts that it "contains a kernel of truth" (1994, p.35).
Although Kuhn himself originally referred to paradigms as incommensurable (1970 {1962} , p.103, p.112, p.157), he objected to the interpretation that this meant there could be no communication between scientists working in different paradigms, and argued that he had clearly implied that communication would be "only partial" (1970 {1969} , p.198). Kuhn later reformulated this "incommensurability of viewpoints and … partial breakdown of communication between the proponents of different theories" in terms of people speaking "different languages". For Kuhn these different languages expressed different cognitive commitments, and were suitable for different worlds (1977, pp.xxii-xxiii). Kuhn pointed out that different languages divide up the world in different ways, and so that translation between the languages (and therefore worlds) of two speakers would inevitably involve some change in meaning (1970 {1965} , p.268): thus the ability to grasp another's viewpoint is not absolute, but "inevitably limited by the imperfections of the process of translation and of reference determination" (1977, p.xxiii). Nevertheless, from Bruner's perspective two participants from different backgrounds can move towards co-constructing a dialogue through their attempts to converse.
Indeed the process of undertaking research interviews, and analysing the data, described in this thesis (chapters 4 and 5) is primarily about overcoming these limitations of the 'imperfections of the process of translation and of reference determination'. But as Bruner points out: in dialogue, language is used to provide constant transactional calibration, so it will be assumed that to some extent my colearners and I (as experienced language users) have developed skills in interpreting the 'worlds' of others. Or as Polanyi put it, "to speak is to contrive signs, to observe their fitness, and to interpret their alternative relations" (1962 {1958} , p.82, my emphasis).
From Popper's (1994) perspective, although conversation may be easier between two people who share common 'frameworks' of assumptions, it is potentially less fruitful in the sense of how much the discussants can learn from the dialogue (p. 35). This is an important point for the present research where the colearners and myself started from the perspective that their understanding of chemistry was different to mine, and both parties thought it would be useful to learn more about this difference. That I was aware of the literature about alternative conceptions, and something of Kuhn's notion of paradigms, where my colearners probably perceived the context of the interviews in terms of 'deficiencies' in their knowledge compared to mine, does not detract from this. Both parties to an interview accepted that there were different understandings to explore, and were committed to using language to undertake the exploration. This is part of the rationale of referring to my interviewees as colearners in the research (§4.3.2).
Bruner is known for his work on developing the theory of instruction, and in particular for proposing the importance of 'guided discovery' (Fox, 1993, p.181). Bruner believed that the cultural invention of schooling led to new ways of thinking (Carraher et al., 1991, p.234; Wood, 1988, p.15, p.84; Wood, 1991, p.97), and that schools should be primarily concerned with developing thinking and problem- solving skills in the academic disciplines rather than imparting specific knowledge (Fox, 1993, p.182; Wood, 1988, p.8, p.136). Bruner (1979 {1962} , p.87) believed that discovery methods encouraged children to become constructivists and effective learners.
Bruner describes three levels of representing the world (Bruner, 1977 {1964} , p.208; see also for example Brown, 1977, p.74; Fox, 1993, p.182): the enactive level (through action), the iconic level (through mental imagery) and the symbolic level (through the manipulation of symbols), and he suggests that teaching that starts with the symbolic level will lead to rote learning (c.f. §2.2.5). Bruner's ideas may be seen to draw upon both Piaget (§2.2.1) in that we create knowledge by active restructuring of our experience of the environment (Child, 1986, p.110; Wood, 1988, p.183), and Vygotsky (§2.2.2), with knowledge presented at the symbolic level being akin to scientific concepts that need to be integrated with spontaneous concepts. Bruner has recommended the use of a 'spiral curriculum' where the same material is met at increasing levels of difficulty during a learner's school years (Child, 1986, p.363).
The research reported in the present thesis does not explore the formal teaching that learners received during their A level course, but Bruner's ideas are considered significant to the collection of research data. Bruner has explored the Vygotskian notion of the Z.P.D., and his group suggested the notion of scaffolding (Tharp and Gallimore, 1991, p.48) whereby a learner is guided by an adult, but the degree of support is reduced as the learner is gradually able to undertake the task without assistance. The teacher acts as "a vicarious form of consciousness until such a time as the learner is able to master his own action through his own consciousness and control" so that the learner can "internalise external knowledge and convert it into a tool for conscious control" (Bruner, quoted in Meadows, 1993, p.248). Bruner introduced the 'hand-over' principle: that the learner moves from being an observer to a participant (Rogoff, 1991, p.78; Tharp and Gallimore, 1991, p.50),
"One sets the game, provides a scaffold to assure that the child's ineptitudes can be rescued by appropriate intervention, and then removes the scaffold part by part as the reciprocal structure can stand on its own."
Bruner, quoted in Wood, 1991, p.109
One of Bruner's coworkers has suggested that "it is hard to find problems that are impossible for a child, given some coaching and some external aids" (Olson, quoted in Brown, 1977, p.78), and another emphasises that "built well, such scaffolds help children to learn how to achieve heights that they cannot scale alone" (Wood, 1988, p.80). Although the details of the scaffolding process will vary, the principle is considered to be applicable to learners across a wide age range (Wood, 1991, p.110).
The research interviews undertaken in the present research used probing questions that might well have acted as scaffolding to allow my colearners to move beyond explanations that they could have produced spontaneously. The methodological consequences are discussed in chapter 4 (§4.10.4) – where it is considered that from a Vygotskian perspective it is quite appropriate to explore the students' Z.P.D.s.
§2.2.4: Kelly
Whereas Piaget demonstrated how children's thinking could be so different to that of adults, and Vygotsky emphasised the importance of language as a mediator that led to convergence of understandings within a culture, Kelly's work highlighted how each person's cognitive structure is distinct.
Kelly built a theory of personality that he called 'personal construct theory' (P.C.T.) He described his position as 'constructive alternativism', which emphasised the way an individual's knowledge was provisional (Kelly, 1963, p.15). Kelly's system assumed learning was on-going and central to personality (p.75). A person's way of relating to their world could be understood in terms of their personal construct system. Constructs were not to be seen as so different from concepts, (pp.69-70), but Kelly thought it was productive to construe them as bipolar or dichotomous (Bannister and Fransella, 1986, p.12; Kelly, 1963, p.59), that is, as the basis for making discriminations (Bannister and Fransella, 1986, p.21).
It would seem that Kelly's constructs were not limited to Vygotsky's 'scientific' concepts where words may be used as tools, but also included 'spontaneous' concepts (c.f. Kelly, 1963, p.92) which were not labelled by words or other symbols (p.110).
Driver and Easley (1978), and Gilbert and Swift (1985) have in different ways attempted to define the constructivist movement in science education in contrast to aspects of Piagetian research; and the former authors have also emphasised the significance of Piaget's methodology for (what in 1978 was) the new tradition of investigating 'alternative frameworks' (p.62). However it is the ideas of Kelly which have been used most explicitly as the theoretical basis of the field of enquiry. This is largely due to the work of the 'Personal Construction of Knowledge' group at Surrey University (Gilbert, Pope, Watts and others) who used Kelly's Personal Construct Theory as the basis of their position on the nature of concepts and concept understanding (Watts et al., 1982, abstract). Concepts would not be viewed as logically organised and tightly defined as in a text book, but to be more ephemeral (or perhaps in a currently in vogue term, 'fuzzy') akin to a "fund of expectations" (p.8.) and "the adoption of a point of view" (p.9) (see §2.10.2, below).
Kelly's work was even more significant for providing the epistemological position of 'constructive alternativism', that individuals construct models of their environment, based on tentative hypotheses, which are tested against experience and modified as required (Pope and Gilbert, 1983, p.196-7).
Kelly's central metaphor was of man-the-scientist (1963 {1955} , p.4), and this was reflected in Driver's focus on the pupil-as-scientist (Driver and Erickson, 1983; Driver, 1983). Driver referred to pupils' construct systems as "spectacles of their own preconceptions" which – in my terms – may act as substantive learning impediments. That is, that many learners "have difficulty in making the journey from their own intuitions to the ideas presented in science lessons" (1983, from the preface). She compared this 'journey' to "paradigm shifts in their thinking" (p.9) and noted how some of the children's ideas were similar to historical scientific ideas (p. 76), something that had previously been noted by Piaget (Driver and Easley, 1978, pp.75-76).
An important aspect of Kelly's theory for the present research is that it considered the issue, discussed above (§2.2.4), of how communication can occur between people holding disparate models of the world. Kelly thought that differences in the construct systems of two individuals could be overcome by the ability of individuals to 'construe the construct system of the other' (Pope and Gilbert, 1983, p.197). In other words, one may develop one's construct system to include a model of the other's version of the world: it is possible to build into one's cognitive structure not only models of how one construes the world, but models of how one construes others to model the world. Examples might include scientists working in two different Kuhnian paradigms, a historian trying to understand the phlogiston theory, or a science teacher trying to make sense of his learners' ideas.
Despite the emphasis on Kelly's ideas, his methodology has not tended to feature in research in learners' ideas. In chapter 4 I discuss why this is, and how I incorporated some of Kelly's methodology – the construct repertory test – into my own research (§4.7).
§2.2.5: Ausubel
Another psychologist who influenced the A.C.M. was Ausubel, who was known for his adage that one should find out what a learner knows, and teach accordingly. Ausubel introduced the notion that learning needed to be meaningful (c.f. rote), and this depended on the learner's cognitive structure, and the nature of the material to be learned (Ausubel, 1961, p.18; Ausubel and Robinson, 1969, pp.50-51). Ausubel and Robinson suggest three conditions for meaningful learning to occur:
- (a) The material itself must be relatable to some hypothetical cognitive structure in a nonarbitrary and substantive fashion.
- (b) The learner must possess relevant ideas to which to relate the material.
- (c) The learner must possess the intent to relate these ideas to cognitive structure in a nonarbitrary and substantive fashion.
(Ausubel and Robinson, 1969, p.53.)
Whilst the third item, which relates to motivation and metacognition, is important, it is the former two points which are of most concern here (c.f. §1.5). Firstly if new material presented to a learner is not able to be related to his or her existing knowledge (i.e. the conditions for a null learning impediment as discussed in chapter 1, §1.5.2) it can not be learnt in a meaningful way, and must be learnt by rote if it is to be learnt at all. In Ausubel's terms, the learner is not "able to effectively exploit his existing knowledge as an ideational and organizational matrix for the incorporation, understanding, and fixation of large bodies of new ideas" (Ausubel and Robinson, 1969, p.57). Secondly, if meaningful learning implies some form of integration or assimilation with existing knowledge, then if the existing knowledge is contrary to that of the authority (teacher, textbook etc.) the new material will not be understood in the way intended (i.e., in the terms presented in chapter 1, there will be a substantive learning impediment).
Such a view leads to an approach to teaching that emphasises the importance of, and seeks to build upon, the learner's existing ideas; rather than marginalise them and assume that they will be overwritten by instruction. It was this perspective that led to the development of the field of research in which this study is located.
§2.3: Children’s science
A learner's existing cognitive structure may therefore be considered a major variable in any learning episode. The A.C.M. developed as a research programme concerned with investigating what ideas learners were bringing to science classes.
§2.3.1 Preconceptions, misconceptions and alternative frameworks
It is appropriate that Driver's words should stand at the head of this chapter, as one of the seminal papers in the field was published by Driver and Easley in 1978. Their paper began with an extract from two fourteen year old pupils discussing thermal expansion, and referring to the molecules expanding, and to the 'heat molecules' (c.f. §3.1.2). Driver and Easley asked about the status of such statements: whether they were "misconceptions, errors, partial understandings or misunderstandings?" (p.61).
They make a reference to the ideas of Ausubel (see above, §2.5.5), who would label such ideas as 'preconceptions', but they felt that the term did not acknowledge how such notions could have the status of models and theories (p.62). The alternative term 'misconception' implied a misunderstanding of formally taught material, and excluded intuitive theories – "the situation in which pupils have developed autonomous frameworks for conceptualising their experience of the physical world" (p.62): i.e. in my terms it recognised epistemological learningi impediments but not ontological learning impediments (c.f. §1.5.3-5). They suggested instead the more inclusive term alternative !ameworks. Alternative frameworks referred to conceptual frameworks which led to the accepted science being counter intuitive, or significantly different to the learner's ideas (Driver, 1983, p.3).
Pope and Gilbert recognised that the notion of 'alternative frameworks' could be related closely to Kelly's P.C.T. (see above, §2.2.4), where each learner held a unique, and dynamic system of personal constructs (Pope and Gilbert, 1983, p.197).
However, as will be discussed below (§2.4.1), the absence of agreed terminology in the field has led to misunderstandings about exactly what different workers mean by terms such as 'alternative frameworks'.
§2.3.2: The notion of children’s science
At the time of writing their paper Driver and Easley were able to describe naturalistic studies based on clinical interviews as "usually small scale and scattered" (1978, p.77). However by 1983 Driver and Erickson referred to a "growing interest" in this work (p.38-39, see also Gilbert and Watts, 1983, p.61). For example the Learning in Science Project which ran from 1979 to 1984 at the University of Waikato, New Zealand (Osborne, 1980), and work based at the University of Surrey made extensive use of these techniques.
One such study was a 1982 paper by Gilbert (Surrey, U.K.), Osborne (Waikato, N.Z.) and Fensham (Monash, Australia) which prominently used the term 'children's science' to describe the "conceptual structures" which children used to understand the world prior to formal instruction (p.623, p.627), and could therefore lead to what I have labelled ontological learning impediments.
Gilbert et al. contrasted children's science with scientists' science – the consensual scientific view of the world and meaning for words (p.627) – and with teachers' science – which was different again, but could usually be considered to fall somewhere between the two (pp.627-8). Gilbert et al. also recognised that the 'viewpoint presented' in science classes matched none of these, and was the result of teachers' science being mediated through the presentation of the curriculum (p. 628). This 'viewpoint presented', which is the teacher's interpretation of the curriculum in view of his or her own understanding of science, I will for consistency refer to as curriculum science .
§2.3.3: The range of children’s science
Research soon demonstrated that children brought their own alternative ideas to the classroom relating to most, if not all, areas of the science curriculum (Driver, 1983, p.7). Driver directed the Children's Learning in Science Research Group (CLiS Project) at Leeds which set about investigating children 's science in a range of curriculum topics (as well as developing constructivist teaching schemes for some), and in 1994 Driver and coworkers published a review of research into children's ideas which covered many aspects of secondary school science.
§2.3.4: The implications of children’s science
In 1982 Nussbaum and Novick reviewed the 'numerous reports' on alternative frameworks and concluded that almost all suggested that such frameworks
interfered with intended learning (p.184). Ault, Novak and Gowin (1984), studied learners' notions of the molecule concept working with the same individuals on two occasions (in second grade, and then in seventh grade) and concluded from their study that the differences in the conceptions in the early grade were significant for later understanding, and that the learners' meanings as grasped in primary grades would effect their understanding years later (p.459).
Given that research suggests that the learner's existing cognitive structure has a major influence on what is learned, and how it is learnt, it becomes important for teachers to know about, and take account of, the ideas learners bring to class. Driver and Erickson set out their empirical premises in 1983 (p.39):
- Many students have constructed from previous physical and linguistic experience frameworks which can be used to interpret some of the natural phenomena which they study forma$y in school science classes.
- These student frameworks often result in conceptual confusion as they lead to different predictions and explanations !om those sanctioned by school science.
- Well-planned instruction employing teaching strategies which take account of student frameworks will result in the development of frameworks that conform more closely to school science.
Gilbert et al. (1982) considered the possible outcomes of the interaction of children's science with curriculum science. There is a spectrum of possibilities. At one extreme the learners' ideas may be readily displaced by exposure to teaching. At the opposite pole the student frameworks might be so stable that they were completely unaffected by teaching. Either of these possibilities would reduce the whole issue to a purely academic one rather than a matter of serious practical concern for science teachers. (It may also be pointed out that in evolutionary terms it would not be expected that a learner with tota$y labile or tota$y stable cognitive structure would survive to evolve under natural selection: in one case the individual could never learn from new experience, and in the other case the individual would have no stable basis for predicting the future, or planning actions. We would expect evolution to result in the selection of some intermediate situation that was optimal in the environmental conditions where selection operated).
Gilbert et al. discussed the outcomes possible in teaching situations (1982, p.630):
- Sometimes there was a 'unified scientific outcome', where the learned meanings closely matched that intended (pp.630-1).
- More often there would be a two perspectives outcome (§2.3.5), where the preexisting conceptions and the newly learnt material would co-exist (p.
- 624).
- Sometimes children's science would be largely undisturbed by 'teaching' (Pope and Gilbert, 1983, p.201).
- There could even be a reinforced outcome (§2.3.6) where the material presented is (mis)understood to support the learner's existing ideas.
- Sometimes there would be partial learning of ideas (§2.3.7), as only so much new material could be learnt at one time (c.f. §1.8), so that ideas would not be fully integrated in cognitive structure, and could be contradictory.
Some of these possibilities are worth further comment:
§2.3.5: The two perspectives outcome
Gilbert and coworkers reported that often there would be a two perspectives outcome, which they described as where "the learned amalgam of children's science and teacher's science can co-exist", so students could be successful in school tests whilst retaining their children's science for informal use (1982, p.624). In this situation curriculum science would be effectively rejected for use as a personal model of the world (Pope and Gilbert, 1983, p.199). One might relate this to the notion of a fragmentation learning impediment (§1.5.2) where the learner does not recognise that existing ideas (children's science) are related to material presented in science classes.
Driver and Erickson (1983) suggested that that whereas scientists have to closely relate their formal conceptual knowledge to their experiences of the world, school children did not generally demonstrate such integration within cognitive structure (p.46). They reported that the two different domains of knowledge could be elicited by different modes of data collection.
Studies showed that learners were more likely to apply scientific principles if questions were set as formal exercises with obviously 'scientific contexts', but they tended to revert to using their alternative frameworks in novel – and particularly 'everyday' contexts (Bliss et al., 1988; Driver, 1983, p.38, p.70; Dumbrill and Birley; 1987; Viennot, 1979, 1985a, p.433). The tendency to pay heed to irrelevant contextual factors in questions can decrease with age, but even University students can change their reasoning in (scientifically) similar questions due to perceived contextual cues (Palmer, 1997).
The notion that a learner could hold distinct, and perhaps contradictory, conceptual schemes for a single topic area has led to considerable debate about the nature – and particularly the status – of children's science. Some workers have argued that children's alternative ideas do not have the form of conceptual frameworks (§2.5, §2.6), or that their ideas make up part of a system of knowledge that can not be classed as 'science' (§.2.7). These criticisms of the A.C.M. are considered in detail below.
§2.3.6: The reinforced outcome
Sometimes instruction would result in a reinforced outcome where the material presented is understood to support the learner's existing ideas, so that for example new terms are used to label existing ideas (Ault, Novak and Gowin, 1984, p.459). Driver has also pointed out that in practical work conceptual frameworks may "restrict empirical observations" (1983, p.65, and see p.27, p.35; c.f. Kuhn 1970 {1962), p.79) and there may be attempts to 'save the phenomenon' (Driver, 1983, p.39). In these cases the children's science acts as substantive learning impediments (§1.5.3) to the intended learning.
§2.3.7: Partial learning of presented ideas
Sometimes instruction would result in partial learning of ideas. As de Bono (1969) has pointed out the human brain is generally not very efficient at accurate precise recall, but rather excels at processing data in the light of existing cognitive structure (p.22). Where existing cognitive structure acts as a substantive learning impediment (§1.5.3) then although changes in cognitive structure would take place, the learner would reinterpret the presented curriculum science in terms of his or her children's science and the 'learnt' version would be very different from that intended by the teacher (Ault, Novak and Gowin, 1984, p.459). In such a context Driver and coworkers refer to Vosniadou and Brewer's idea of assimilatory concepts (§2.3.10), described as "attempts on the part of children to reconcile their presuppositions… with the information they receive from the adult culture" (quoted in Driver et al., 1994b, p.87).
§2.3.8: Children’s science and the nature of the curriculum
Gilbert and his coworkers then proposed a range of possible outcomes when a learner possessing existing alternative ideas about a topic was presented with formal instruction. Driver and Erickson (1983) thought that the actual outcome was likely to vary with the nature of the taught material, so that stable alternative frameworks were more likely to interfere with learning of topics where learners will have rich early experiences (such as heat, mechanics, light) (p.49). Chemical bonding would not be considered such a topic (c.f. §1.3.2). (In terms of the notion of learning impediments used in chapter 1, we might expect teaching about chemical bonding to be impeded less by ontological learning impediments than teaching about topics such as mechanics.) Any alternative conceptions about chemical bonding might therefore be expected to be relatively labile. However, Driver and Erickson recognised that even in such curriculum areas not closely related to everyday experience a learner was likely to draw analogies with existing knowledge to make sense of the topic area (p.49, c.f. §2.8.4).
§2.3.9: The status of children’s science: barriers or bridges?
The lack of understanding about how conceptual development occurs (see §2.10) has resulted in uncertainty over how learner's alternative ideas should be considered. Researchers who see children's alternative ideas in science in terms of misconceptions would view these ideas as barriers to 'correct' learning. The A.C.M. have taken an alternative view: that as children's science is the starting point from which further learning must occur, it is appropriate to consider children's science as 'bridges' or 'stepping stones' on the 'path' to the intended understandings of curriculum science.
The constructivist writers recommended that teachers should develop diagnostic skills to identify children's science, (Watts et al., 1982, p.3, p.27). They presented a vision of science teaching as building on the foundations of children's ideas, and developing and extending their thinking towards the scientific models (Driver, 1983, p.3; Gilbert et al., 1982, p.631; Watts et al., 1982, p.7, p.27), and they have emphasised the potential links between children's science and curriculum science (Ault, et al., 1984, p.459). More recently, the influence of the Vygotskyan perspective (see §2.4.2) may be detected in the work of writers who see learning as a sociocultural process of bridge-building between informal and formal knowledge systems (Driver et al., 1994b, p.94; Hennessy, 1993, p.7).
In emphasising the importance of children's science the constructivists provided ammunition for critics such as Matthews (1993) to accuse them of being 'relativists' (c.f. §2.0). Examples include:
- teachers should not think of "a matter of not understanding but of understanding differently from what was intended" (Nussbaum and Novick, 1982, p.184);
- it was necessary "to explore and empathise with children's frameworks" (Watts, et al., 1982, p.27), because they had "both important epistemological value and high educational status" (p.7);
- learners' ideas are "personally viable constructive alternatives – rather than the result of some cognitive deficiency, inadequate learning, 'carelessness' or poor teaching" (Gilbert and Watts, 1983, p.67);
- alternative frameworks uncovered in Watts' research were described as "coherent, interna$y logical conceptual frameworks based upon [pupils'] own experiences which are very successful in explaining everyday events", and it was suggested that they should be given "due status" (Pope and Gilbert, 1983, p.198) as they were not only "plausible" but also "!uitful" for the pupils (p.199);
- learners' alternative frameworks were "in keeping with their experience and in this respect … not 'wrong' … perhaps just not as inclusive as the accepted 'scientific' view" (Driver, 1983, pp.87-88).
Although these workers were right to emphasise the pedagogic importance of children's ideas in science, it could be argued that in view of the traditional philosophical outlook of many school teachers (see §2.0), with a realist notion of knowledge (Pope and Gilbert, 1983, p.193; Tobin, 1990, p.34) such 'relativist' statements might have deterred some educators from taking up their ideas. Writing that was intended to 'sell' the constructivist position may have failed to take the existing conceptions of an important part of the audience – classroom practitioners – into account.
§2.3.10: Stepping stones between children’s science and curriculum science
According to Driver, the building of bridges between children's science and curriculum science may involve 'intermediate notions' or 'intermediate conceptions' (Driver, 1989, p.483; Driver et al, 1994b, p.81), so that progression may follow conceptual trajectories, defined as "a sequence of conceptualizations which portray significant steps in the way knowledge within the domain is represented" (Driver et al., 1994b, pp.85).
Driver suggested that one part of the learning process should involve learners' theories being made explicit so that they could be compared and challenged (1983, p.76), and this was indeed a key part of the teaching schemes developed by her CLiSP research group (Brook and Driver, 1986; Wightman et al., 1986). Pope and Gilbert also suggested that learners should be encouraged to reflect on their ideas, and thought that their role as constructors of theory should be explicit (1983, p.193). Ault, Novak and Gowin suggested that learners should learn to spot the 'tangles' and 'twists' in their own conceptual schemes (1984, p.460). Unfortunately more recent research into children's ideas about the nature of science suggests that although they do use models, they often do not generally have the metacognitive awareness to conceptualise their thinking in this way (Driver, et al., 1996, p.139, Duveen et al., 1993, p.25).
§2.3.11: The value of having alternative conceptions
Ault, Novak and Gowin (1984) studied learner's notions of the 'molecule' concept using a method for interpreting and representing data collected through clinical interviews (p.441). As they investigated the same individuals on two occasions (in second grade, and then in seventh grade) they were able to draw some conclusions about the development of conceptual understanding. They found that it was better for a young pupil to have a variety of alternative conceptions than few conceptions at all, as understanding evolved more rapidly from a rich conceptualisation. If a pupil in an early grade held a range of idiosyncratic meanings these would tend to persist, but still provided a better structure for conceptual development, than a poor range of notions (p.459-60).
Hennessy (1993) points out that while some alternative conceptions are best considered as the foundations for building conventional notions, others might be of value in themselves, as the alternative models may be more appropriate in some contexts (p.9). Whilst this may seen to reflect the relativist statements criticised above (§2.3.9), this is certainly a view which could have some value in chemistry with its manifold models (§1.7.1). The 'situated cognition' perspective is considered further below (§2.7.2).
§2.4: The units of analysis of children’s science
In 1988 Abimbola made the seemingly obvious point that "it is important that science education researchers understand themselves when they use terms that describe student conceptions in science" (p.175). The platitude was appropriate in view of lack of consistency of the nomenclature being used in the field. In 1983 the constructivist movement in science education had been described as being in a pre- paradigmatic phase (Gilbert and Watts, 1983, p.61), and in particular it was noted that there was little agreement over the terms used to describe aspects of learners ideas elicited during research (Driver and Erickson, 1983, p.46; Gilbert and Watts, 1983, p.69). The use of term 'frameworks' was especially unclear, and although there were attempts to clarify it, these did not succeed (Black and Lucas, 1993, p.xii).
Gilbert and Watts highlighted the need to distinguish between "an individual's psychological, personal, knowledge structure" (pp.45-65) – e.g. the 'concept' as inferred to be in the learner's head – and aspects of "the organisation of public knowledge systems", i.e. the orthodox academic version of the 'concept' as presented in the textbooks etc. Phillips has criticised the error of confusing these two distinct phenomena, "disciplinary structure and cognitive structure", in the work of Piaget (Phillips, 1987, p.139).
I would suggest that there are a number of other distinctions that need to be kept in mind when talking about learners ideas:
- between an individual's cognitive structure (which can only be conjectured, §1.4.1), and researcher's own models (which can be formally represented in words and diagrams);
- between representations of aspects of an individual's cognitive structure, and general models which are intended to reflect commonalities from the representations derived from several individuals;
- at different scales within an individual's cognitive structure.
The first of these distinctions closely parallels the distinction that Gilbert and Watts, and Phillips, have emphasised, but where the disciplinary structure of interest is that of constructivism itself. Ault, Novak and Gowin (1984) are careful to distinguish cognitive structure from the representation that is a product of their analysis. They use research data to infer a 'conceptual structure' that they consider "a best approximation of cognitive structure, the 'true object' of interest" (1984, p. 446), but this terminology has not been widely taken up. Lakoff and Johnson describe the difference between our cognitive structures and our public representations of these structures as a "most important distinction" (1980, p.206).
§2.4.1: The meaning of the term alternative frameworks
All three distinctions are confused in the various uses of the word 'framework'. Driver clearly used the terms 'conceptual framework' and 'alternative framework' to refer to aspects of an individual learner's cognitive structure, that is for "the mental organisation imposed by an individual" (Driver and Erickson, 1983, p.39) which was utilised "for conceptualising their experience of the physical world" (Driver and Easley, 1978, p.62).
In contrast the Surrey group used the same term to refer to their representations of commonalities in their models of aspects of the cognitive structures of many individuals, so that a framework may be described as "a composite picture based upon ideas shared by a number of pupils" (Watts, Gilbert and Pope, 1982, p.15, my emphasis), "generalised non-individual descriptions" and "thematic interpretations of data, stylised, mild caricatures of the responses made by students" (Gilbert and Watts, 1983, p.69).
To summarise this difference, for Driver, the conceptual framework is part of an individual's cognitive structure, something inside the mind of a learner, where for Gilbert and Watts the 'framework' is something presented in the public domain that is constructed by the researcher on the basis of data collected from a selection of learners.
Gilbert and Watts suggested that the term 'conceptions' "be used to focus on the personalised theorising and hypothesising of individuals" (1983, p.69), although in places they appeared to use the word 'framework' when by their own definitions they meant 'conceptions' (Gilbert and Watts, 1983, p.83, p.86; Watts, 1982, p.116; 1983a, p.217; Watts and Gilbert, 1983, p.168-9). Gilbert and Zylbersztajn (1985) refer to alternative conceptions as "alternative views of the world" and "personal
explanations, which make sense from an individual's point of view". They suggest conceptions may take "the form of expectations, beliefs and meanings for words" (p.107). In practice the term 'conception' seems to have sometimes been used synonymously with 'framework', as when Watts and Gilbert report that "eight conceptions [sic] of the word force have been identified … and seven conceptions [sic] for energy" (Watts and Gilbert, 1983, pp.161). Hewson actually notes that in her writing "alternative frameworks are referred to as alternative conceptions" (1985, p. 154).
One source of confusion is when an author makes references to students using !ameworks. This seems to have two distinct meanings. Firstly learners are said to use their frameworks when perceived regularities in their answers to questions su&est a coherence and logic to their thinking processes, which leads to the hypothesis that the topic being explored is represented in some form in cognitive structure. In other words we are talking about something which is inherent in the learner, and exists before the researcher sets about interpreting data.
At other times the statement that learners use frameworks appears to mean something very different: that once the researcher had set up a model consisting of a set of representations of possible alternative meanings for the focal topic, it was then possible to support the authenticity of the analysis by demonstrating the compatibility of representations with extracts of interview transcript.
§2.4.2: Alternative frameworks may be ubiquitous or elusive: the danger of 'framework spotting'
This distinction is important because although constructivism has been considered a dominant paradigm in science education research (Solomon, 1994, p.7), it does not have universal acceptance amongst active researchers in science education (e.g. Kuiper, 1994). Indeed it is probably more helpful to think of Lakatos' notion of research programmes here (1970 {1965} ).
The research reports of the constructivists (such as Watts 1982, 1983a, 1983b) could be argued to be based around two claims:
• claim 1: learners have 'alternative frameworks1'.
• claim 2: Watts and others have constructed some 'alternative frameworks2'.
where the subscripts refer to two discrete meanings of the term framework: alternative frameworks1 are "the mental organisation imposed by an individual on sensory inputs" (Driver and Erickson, 1983, p.39), and alternative frameworks2 are "thematic interpretations of data, stylised, mild caricatures of the responses" (Gilbert and Watts, 1983). For a worker within the paradigm (Kuhn) – participating in the same research programme (Lakatos) – these two knowledge claims have a very different status.
Claim 1 is a reference to one of the key aspects of the disciplinary matrix (Kuhn). In Lakatos' terms it is part of the hard core of the research programme: something that is protected by the negative heuristic, i.e. workers within the research programme will not find it fruitful to undertake research to falsify the claim, which is an essential prerequisite for the programme to proceed. Claim 1 is a claim about the tenets of constructivism.
Claim 2 concerns only the proficiency of the researcher as a competent practitioner in the field. The positive heuristic of the research programme leads to enquiry into eliciting learner's ideas, and constructing representations which (a) other researchers accept as authentic, and (b) are in a form which may be readily communicated to those teaching science. Individual examples of claim 2 make up part of the protective belt of theory in the research programme.
For researchers in the constructivist tradition claim 1 is therefore axiomatic, whereas specific examples of claim 2 are peripheral. A statement that alternative frameworks1 do not exist is an attack on constructivism, and by definition can not be made from within the research programme. A statement that certain alternative frameworks2 do not exist is merely questioning the work of one individual researcher, and does not necessarily have serious implications for the research programme. Indeed criticisms of alternative frameworks2 would be quite proper within the programme, and such debate would indeed be directed by the positive heuristic. It is analogous to the difference between denying the existence of fairies at the bottom of the garden, and arguing over what their names are. (Or the difference between denying that continental drift takes place, and disagreeing over the precise rate at which the Atlantic is spreading.)
Having established this distinction it is now appropriate to consider an example of criticism of constructivist research – on the theme that alternative frameworks do not actually exist. In 1983(a) Watts published results of his research into student understanding of the concept of force, summarising his findings in terms of eight alternative frameworks. Eleven years later Kuiper (1994) questions the use by some science education researchers of the assignation "alternative frameworks" to aspects of student thinking revealed during enquiry. In Kuiper's own study of student ideas about force he not only failed to support the specific findings of Watts and others, but made a claim that:
- "students in general do not have an 'alternative framework' for force" (p.279).
This discrepancy requires some comment. However a reading of the meanings these two authors give to the term 'framework' is enlightening. Kuiper comments that,
"The use of the term framework in the description of student understanding implies an ordered and schematic understanding of a concept. This term can be understood to mean that a particular student has a set of student ideas concerning one and the same concept which appear to be logically coherent and ordered."
Kuiper, 1994, p.280, my emphasis
There are two important points that need be raised from this extract. The first relates to the basic tenets of constructivism on which much of this field of research is 'built'. According to P.C.T. (Kelly, 1963, see §2.2.4) the learner and the researcher have different construct systems, and that what is "logically coherent and ordered" in terms of the learner's construct system, may well appear to be confused and contradictory when re-interpreted through the researcher's own construct system. The research process may be understood in terms of the researcher developing his or her construct system to try and see things the way the learner does. Kelly refers to "the credulous attitude" (1962, p.174), and the metaphor of putting on different goggles has been used (Pope and Watts, 1988). Kuiper used a test where "the same problem was put four times, only placed in different contexts" (p.281). The assumption being that "if students are to have alternative frameworks, then certainly the same problem, in different contexts, should be answered with the same student idea" (p.281). However it could be argued that what was the same problem for the researcher was several different problems when viewed through the construct system of the learner. As Edwards and Mercer point out, context "is essentially a mental phenomenon", and "participants' conceptions of each other's mental contexts may be wrong or, more likely, only partially right" (1987, pp. 160-161).
So there is the danger of a tautological argument here. The argument could be summarised as what is 'alternative' does not appear consistent and therefore cannot be a '!amework', whilst that which qualifies as a '!amework' must seem logical within the researcher's own construct system, and is therefore not considered 'alternative'. 'Alternative frameworks' therefore become a logical impossibility within the perspective in which the research is conducted. The discussion also ignores (or perhaps excludes) what have been termed 'multiple frameworks' (discussed below, §2.5.2).
The second point of importance here is that Kuiper is referring to individual conceptual frameworks, or alternative frameworks1, whereas Watts' research was reporting alternative frameworks2 – "a simplified description" that came "from no one pupil", but had "been pieced together from the implicit and explicit conceptions used by the children" to "form a composite picture based on the ideas shared by a number of pupils" (1983a, p.218).
These two workers were also using different techniques to collect data. Watts was working with interviews, an approach that allows considerable interaction between the discussants (see for example Powney and Watts, 1987) and has been found particularly useful for eliciting learners' conceptions (§4.6); whereas Kuiper used written test items which do not provide the flexibility of the 'conversational' approach fundamental to constructivist enquiry. Whereas Watts was working within the 'interpretive' research paradigm (i.e. 'paradigm 2' in terms of Gilbert and Watts (1983), see §4.1), Kuiper's paper is clearly based on a very different – i.e. normative – approach ('paradigm 1' according to Gilbert and Watts), using the written responses classed against a small number of predetermined categories.
Pope and Denicolo (1986) had foreseen the dangers of 'frameworks' in the naturalistic research literature being taken out of context by those working in a reductionist manner. They explained that authentic reports would describe the subtleties of individual learners' ideas, and the stages of data reduction used in the analysis. However such reports would be too detailed and dense to be read by the teachers to whom they were addressed, and too long to be published in most journals. Thus "authenticity must be tempered with utility" (p.156). Pope and Denicolo used Watts' work as an example: as he had "clearly described his data degradation process as he moved from consideration of [the] child's alternative conceptions, through categorisation of exemplars of these conceptions using verbatim quotes as evidence to the production of descriptions of a range of alternative conceptual frameworks" (p.157). They warned against ignoring this analytical process ("although starting from a holistic approach one 'end product' of his work is a much reduced description of the construing of the individuals in his study which, if taken out of context, is also devoid of consideration of the particular choices made by the researcher in his conduct of data collection and analysis", p.157), and suggested that,
"the busy teacher or researcher with a predilection towards reductionism may well ignore the 'health warnings' conveyed in our research report. Instead they will indulge in a 'framework spotting' exercise using reified descriptions of frameworks and ignoring the ontology of these frameworks."
Pope and Denicolo, 1986, p.157
Kuiper's study suggests that he has indeed ignored Watts' 'health warnings', and – at least in part – this is the error of reading reports of alternative frameworks2, and assuming they refer to alternative frameworks1.
§2.4.3: Alternative frameworks and alternative conceptions
With regard to the third distinction in my list, Driver included in the term alternative frameworks both "an idiosyncratic response to a particular task" and "general notions applied to a range of situations" (1983, p.7). I would argue that it is useful to draw a distinction between aspects of cognitive structure which influence student behaviour (such as answers to questions) in response to a range of stimuli (such as a series of questions on a topic area), and the level of thinking that produces a single proposition. It has been suggested that "propositions are the 'molecules' from which meaning is built and concepts are the 'atoms' of meaning, to use a rough metaphor" (Novak, 1985, p.192). The identification of conception with proposition seems to reflect workers such as Hewson for whom "the term conception is used to indicate a functional unit of thought" (1985, p.154), and Strike and Posner who "make no distinction between conceptions and ideas and use the terms interchangeably" (Strike and Posner, 1985, p.213, italics in original).
The term framework suggests 'a basic structure which supports and gives shape' (Longman Modern English Dictionary, Watson, 1968) and Abimbola uses !amework in such a sense, as "the organization of ideas rather than the ideas themselves", so that "alternative !ameworks are just the undergirders that anchor ideas" (1988, p.181). This is consistent with Ault and coworkers' notion of conceptual structures, models of aspects of cognitive structure "likely to generate" a student's "claims about events" (Ault et al., 1984, p.446, my emphasis), and to Viennot's approach to investigating within which framework elicited student conceptions occur (Viennot, 1985a, p.433).
§2.4.4: Alternative frameworks and gestalts
The two terms considered above, conceptions and !ameworks, are very commonly used in the literature. The term gestalt is much less common, but may also be useful in explaining learners' ideas in science. A gestalt is an integrated whole, and this idea forms the basis of a school of psychology (e.g. Pearls et al.,1973 {1951} ) that is based on the premise that "mental processes and behaviour cannot be analysed, without remainder, into elementary units, since wholeness and organisation are features of such processes from the start" (Drever, 1964 {1952} , p.108).
Lakoff and Johnson (1980) have argued that the human conceptual system largely functions through metaphor containing metaphorical as well as nonmetaphorical concepts (p.195). They define nonmetaphorical concepts as "those that emerge directly from our experience and are defined in their own terms, [such as] … spatial orientations … ontological concepts … structured experiences and activities" (p.195). M etaphorical concepts "are those which are understood and structured not merely on their own terms, but rather in terms of a different kind of object or experience" (p.195, italics in original). They describe the metaphorical structure as "extremely rich and complex" (p.195).
These authors claim that metaphorical concepts are grounded in experience (p.204) and are "based on complex experiential gestalts" (p.201), by which they mean "a multidimensional structured whole arising naturally within experience" (p.202). A particular experiential gestalt is described as being either
- "a structure within a person's experience that identifies that experience as being of a certain kind", or
- "a structure in terms of which a person understands some external occurrence and that identifies that occurrence as being of a certain kind" (p.205).
Like !ameworks, gestalts may be understood as reflecting an aspect of cognitive structure which is used to interpret perceptions and organise thoughts. However whereas !ameworks may be considered to reflect aspects of cognitive structure that are to a large extent consciously available to the learner – so that it would be possible to sit down and discuss an individual !amework with the learner in terms of the propositions from which it is constructed, and to authenticate the !amework in a piecemeal manner – the gestalt may be envisaged as reflecting an aspect of cognitive structure which is a fundamental aspect the individual's world view, but which is largely tacit.
Perhaps such gestalts (or rather the cognitive structures that lead to their perception) are not unlike Piaget's "system of mental tendencies and predilections of which the child himself has never been consciously aware and of which he never speaks" (1973 {1929} , p.14), and are akin to Vygotsky's spontaneous concepts that "the child becomes conscious of … relatively late" (1986 {1934} , p.192, §2.2.2). Vygotsky discusses the process whereby "scientific concepts … supply structures for the upward development of the child's spontaneous concepts towards consciousness and deliberate use" (p.194). As the structures which lead to the perceptions of gestalts do not concern classes of objects, but broad basic assumptions about the way the world is, it might be expected that the process of bringing them to conscious and deliberate use is more complex than that envisaged by Vygotsky.
Polanyi (1962 {1958} ) has described what he calls the ineffable domain, as "where the tacit predominates to the extent that articulation is virtually impossible" (p.87), and has pointed out that this domain can not be examined by introspection,
"The curious thing is that we have no clear knowledge of what our suppositions are and when we try to formulate them they appear quite unconvincing."
Polanyi, 1962 {1958} , p.59
The distinction between frameworks and gestalts may be useful even if it does not represent an absolute division in cognitive structure. Because the proclivities leading to gestalts concern fundamental ways of organising
experience of the world they can influence thinking over a wider range of phenomena and situations than the framework; and because that are largely tacit, they are more difficult to investigate than 'conceptions' or 'frameworks'. The research literature includes a multitude of alternative conceptions, a range of alternative frameworks, but only a few papers claiming to have uncovered gestalts.
Anderson (1986) proposed that one of Lakoff and Johnson's gestalts, the 'experiential gestalt of causation' was of particular significance: "a common core to … pupils' explanations and predictions in such widely differing areas as temperature and heat, electricity, optics and mechanics" (p.155). This gestalt concerned causality, and the suggestion was that many disparate phenomena were perceived in terms of an object that is acted upon by an agent through the use of an instrument. The suggestion was that where conventional science teaching concerned phenomena that were not presented in these terms, misunderstanding could commonly occur. For example, in the case of inertia (i.e. Newton's first law of motion, §3.1.3): "the idea of motion with no force whatsoever goes absolutely against the experiential gestalt of causation, that successful organizer of so much experience" (p.169). The tacit nature of the gestalt may be understood in terms of it developing in the very young mind before the acquisition of language.
Watts and I have suggested that data from learners in a similarly wide range of science topics supports the existence of another gestalt that forms the basis for deciding which phenomena need to be explained in analytical terms, and which are accepted as just being. This explanatory gestalt of essence (Watts and Taber, 1997) is proposed to explain why it is that learners are often prepared to give reasons – sometimes quite creative ad hoc reasons – to explain many phenomena, but for others seem nonplused by the question and can only respond in terms of "it's natural". This response seems unlikely to mean just 'I don't know and I can't think of any suggestions' when the same individuals show fertile imaginations in response to different questions. It seems that learners genuinely construct explanations for phenomena that are recognised as justifying them – both spontaneously and when provoked by educational researchers – but reach what for them are 'first principles' where the idea of further explanation becomes non-sensible.
Again it seems likely that this gestalt has its origins in early life experiences, so that the learner is not consciously aware that he or she reaches a point where explanation comes to a stop, and so when these 'first principles' are questioned it is not just that the learner can not answer the question, but rather that he or she perceives the question as meaningless within his or her world view (akin to the cosmologist frustrated by the question "if time and space were created in a 'big bang', then what existed before this?")
The experiential gestalt of causality is proposed as a representation of something universal in human cognitive structure. Similarly, the explanatory gestalt of essence seems to be widespread: learners generally appreciate the function of explanation – but often seem to reach the "it just is" point.
§2.4.5: Phenomenological primitives
The notion of phenomenological primitives, or p–prims, may be considered a similar idea to that of gestalts. Hammer (1996) describes how diSessa's p–prims are considered to be stable aspects of cognitive structure, which make up one's intuitive physics. These primitives are small units, such as closer means stronger (p.7) which are potentially widely applicable. From this perspective many alternative conceptions elicited from students are not in themselves necessarily stable, but are constructed in situ using one of the p–prims held in cognitive structure (so once a 'closer means stronger' p–prim is developed it might be applied in a wide range of contexts such as loudness, intensity of light, and so on).
§2.5: Criticisms of the notion of alternative frameworks
"One open question is the extent to which children's conceptions are genuinely 'theory- like', that is having a coherent internal structure and being used consistently in different contexts…"
Driver, 1989, p.483
The constructivist position outlined so far might be summarised that
- learners, especially children, come to science classes with their own explanations of scientific phenomena, and that
- this children's science may interfere with the intended learning.
This position has been widely debated and point 1 is generally accepted. However there has been considerable debate over the nature of such children's science. Kuiper's criticisms of the the idea of alternative frameworks have been considered above (2.4.2), but other workers have suggested that the A.C.M. may be over- emphasising the extent to which children's science acts as a competitor to curriculum science (§§2.5-2.7).
In particular there have been discussions on
- . the extent to which learners alternative ideas are stable, rather than being largely created in the context of clinical interviews, test situations or social chit-chat;
- the extent to which children's science is theory-like, in terms of having the coherence expected of scientific explanatory frameworks;
- the extent to which children's science comprises of ideas which are integrated together in cognitive structure, rather than being a collection of discrete conceptions.
These debates are very important for the present thesis. If learners hold alternative conceptual frameworks for science concepts which are stable over long periods of time, which are coherent and self-consistent, and which are closely integrated with other concepts represented in cognitive structure, then these might be expected to make up significant substantive learning impediments (see §.1.5.3). If this is the case then the study of such conceptual frameworks might produce much of pedagogic value. Part of the raison d'être for research into learners' alternative conceptions and frameworks is the claim that they can be extremely tenacious and difficult to extinguish (e.g. Driver et al. 1985b) and therefore teachers need to be aware of their existence.
However, if it were to transpire that children's science is a labile collection of incoherent and self-contradictory notions, largely generated through the process of elicitation itself, then it would not be expected to act as substantive learning impediments. (Indeed, as Solomon has pointed out, activities designed to elicit alternative conceptions may catalyse their generation, see §2.8.3). If this were the case then learning difficulties would have to be sought elsewhere (for example as null learning impediments, see §1.5.3).
In practice, a wide range of positions have been taken by workers in the field, all supposedly based on empirical data collected from learners.
§2.5.1: Stability of children’s science
In 1983 Driver reported that research indicated that alternative frameworks did not seen to be extinguished by teaching (p.76). Driver explained that where learners were presented with material at odds with their cognitive structures they had to both understand the new ideas, and to be prepared to move outside of their existing modes of thinking – "to make the intellectual leap of possibly abandoning an alternative framework which until that time had worked well for them" (p.9, see also p.41). Therefore the time-scale over which substantial learning could be expected to occur would be long term, i.e. months and years (Driver and Erickson, 1983, p.54). From a Vygotskian perspective (§2.2.2)
"learning science should involve the gradual integration of personal experience and knowledge into the complex systems of models and theories, and the ways of thinking, that scientists use to explain natural phenomena … children need time to get used to and accept new ideas and other ways of understanding phenomena … and … time to move back and forth between everyday concepts and scientific concepts"
Howe, 1996, pp.47-48
As Hennessy points out scientists become experts through a long process of cognitive apprenticeship, where they spend years acquiring both intuitive knowledge and "sophisticated mental models" of their specialist field (Hennessy, 1993, p.1). (Models of how such change occurs are discussed later in this chapter, §2.10.3. A basic assumption of these models is that conceptual change may be a rational process, which ignores the possibility that learners have have emotional commitment to their existing ideas – particularly where they have acquired them from a significant other such as a parent, close friend or favourite teacher).
Aware that much of the research data available had been collected from individuals on one occasion, Gilbert and Watts suggested the need for "successive re-inquiries" into the frameworks used by individuals over several years (p.87). The present research has such a longitudinal nature (§1.8, and in particular see the case studies in chapters 7 and 8).
Some critics have claimed that there is evidence from many learners that they do not have stable alternative conceptual fr!ameworks at all, but that their informal knowledge is changing all the time. Solomon (see below, §2.7) has discussed Pine's doctoral research into young children's ideas in a range of science topics which suggested that children's notions changed over time as well as being multifaceted and dependent upon context (1992, p.28). She has also pointed out how research into various topics had demonstrated how children seemed to quite readily move between applying different alternative frameworks, even in a single interview (p.24). Other authors (such as Claxton, see below, §2.6) have suggested that student thinking about a topic often seems not so much to reflect a conceptual 'framework' as a wide set of distinct ideas that are applied locally, according to the perceived context. Although workers such as Solomon and Claxton draw attention to an important feature of learners' thinking, it will be argued in this thesis that their criticisms of the notions of stable frameworks are misguided. Firstly it is important to distinguish stability from coherence. This distinction is recognised in Kelly's P.C.T. (§2.4.4) in terms of his 'fragmentation corollary', that is, that "a person may successively employ a variety of construction subsystems which are inferentially incompatible with each other" (1963 (1955), p.83). From a Kellyan perspective, an individual could have a stable construct system, yet give the impression of flitting from one notion to another. The case for learners' alternative ideas being stable does not require learners to have conceptual frameworks which satisfy the criteria applied to public science (in terms of logical coherence). Driver et al. have emphasised the distinction,
"It is often noticed that even after being taught, students have not modified their ideas in spite of attempts by a teacher to challenge them by offering counter-evidence. … Although students' notions may be persistent, as we have already argued, this does not mean the student has a completely coherent model of the phenomena presented. The students' interpretations and conceptions are often contradictory, but none the less stable."
Driver et al., 1985b, pp.3-4
The phenomena of 'multiple frameworks' will be discussed further below in the next subsection (§2.5.2). It should also be pointed out that those researchers who argue that alternative conceptions and frameworks are 'stable' do not consider them to be totally immutable, or there would be little purpose in advising teachers how to bring about conceptual change (§2.10.5).
§2.5.2: The coherence of children's science – multiple frameworks
Some workers have claimed then that children can hold alternative frameworks about scientific topics which are stable, theory-like, and may be consistently applied. However, research into learners' alternative frameworks in different areas of the science curriculum has suggested that the same individual often exhibits evidence of holding to more than one alternative framework for a particular science concept.
Pope and Denicolo (1986) have observed that where researchers had presented a range of alternative frameworks to describe learners ideas on a topic, the data may suggest that some learners exhibited multiple frameworks, that is "where, within one utterance or short speech act, more than one of [the proposed] frameworks was projected" (p.158). Pope and Denicolo suggested that although it would often be possible to artificially 'disaggregate' a learner's statements into smaller parts which could independently be fitted to the different frameworks, such a process was not an authentic representation of utterances that seemed to genuinely encompass several categories that the analyst considers distinct (p.159) .This raises the question as to whether the learner holds multiple frameworks for a topic, or a single coherent framework which does not fit the categories in the researcher's model (which is an abstracted and simplified set of alternative frameworks, see §2.4.2). For as Viennot has pointed out, the very things that make an individual's thinking 'alternative' make it difficult to comprehend, and describe, as the learner may use different terminology, and her alternative notions do not necessarily match the concepts of science through a "one-to-one correspondence" (1985a, p.433. See also my earlier comments about Kuiper's criticisms of alternative frameworks, §2.4.2). The tendency is for the learner's comments to be interpreted through the construct system of the researcher, that is to try and make sense of the learner's utterances in terms of curriculum science (p.433). It would certainly seem feasible that in some cases researchers may have failed to fully appreciated learners' unified alternative frameworks and misidentified utterances as representing several discrete frameworks.
Indeed it might be asked how a researcher could ever understand a learner's thinking if their construct systems were so different. Followers of Kelly might refer to his adoption of the credulous attitude (1963 {1955} , p.174) – perhaps this could be crudely paraphrased that nothing should be ruled out in advance. For Kelly, understanding another person requires the researcher to "subsume the constructs of the subject" (p.174). From the perspective of P.C.T. the researcher has to develop his or her construct system to include new constructs that model the construct system of the learner (see §2.4.4 above), and Kelly warns against "ignoring the personal construction of the [researcher] who does the observing" (p.174).
It is a tacit assumption in the field that – for example – a researcher could acquire a new notion of force, that is inconsistent with his or her own understanding, but which stands for a learner's apparent understanding – and that the researcher can do this without having to give up his or her own alternative (in the literal sense) notion. 'My understanding of learner X's concept of force' may be stored in cognitive structure without significantly interfering with 'my understanding of force'. In a similar way, a historian of science would be able to hold a range of 'versions' of the force concept, which could be labelled 'my understanding of Aristotle's notion of force', '…Galileo's notion…', '…Newton's notion…', and so forth.
It would be generally accepted therefore that an individual could hold 'multiple frameworks' for 'a concept', where these frameworks were understood by the individual to refer to distinct 'versions' of the concept, and therefore actually different concepts. In other words, where the construer has reason to construe distinct versions of a concept, we would expect to be able to infer 'multiple frameworks' within the concept area. In the examples given above (the educational researcher, the historian of ideas) the distinctions between possible frameworks are clear to the conceiver and to others.
If this is accepted then it seems reasonable to suggest that there may be occasions when a learner may perceive (consciously or otherwise) a distinction that justifies applying different frameworks, although the observer may not initially appreciate the distinction. There are reports in the literature that learners may compartmentalise learning in ways that teachers do not intend. For example a learner may effectively master the scientific version of a concept, and apply it in the context of classroom and examination questions, but chose to answer questions in an 'everyday' setting according to an alternative set of ideas (see §2.3.5 above). The learner distinguishes 'life-world' and 'school' knowledge in a way that the teacher does not. (The term 'life-world' is that used by Solomon, as discussed below, §2.7.1). If the learner has to be able to converse with peers and parents in out of school contexts, it way well be more appropriate if 'life-world knowledge' is applied in such contexts, as this is how effective communication will occur. As Driver et al. comment, "human beings take part in multiple parallel communities of discourse, each with its specific practices and purposes" so "we would not expect students necessarily to abandon their common-sense ideas as a result of science instruction" (1994c, p.8).
It would seem then that when a learner seems to display 'multiple frameworks' for a topic it may be that a unified coherent conceptual scheme is not apparent to the researcher because he or she has not been able to fully construe (model) the learner's constructs system. Alternatively, there is no reason to rule out the possibility that learners do indeed hold multiple frameworks to explain some phenomena, (and that they may be triggered into applying particular frameworks by particular cue or mind sets). This idea is developed in the section on conceptual change (§2.10).
§2.6: Claxton’s alternative perspective: minitheories
Claxton (1993) has been one critic who has suggested that it is naïve (a "gross simplification") to infer that learners have alternative frameworks based on the utterances collected during research (p.45). Rather, Claxton points out, such utterances may be reflections of specific circumstances ("an unprecedented question … a unique nexus of opportunities, abilities, constraints and personal history") as much as underlying cognitive structure (p.45). He suggests that our interactions with learners allow us to access their thinking, but that this is a process of constructing ideas in situ, rather than a reporting of stable conceptual structure, that is that the researcher's construction of an 'alternative framework' may be no more than an "ephemeral reflection" of that construction process (p.45). As Ault, Novak and Gowin point out, people may indeed hold "multiple, contradictory notions" but some of these elicited in research may be "transitory artifacts" of the interview itself (1984, p.447).
In Claxton's view, the underlying cognitive structure itself may be better modelled as a large number of discrete 'minitheories', rather than as expansive conceptual frameworks (i.e. learners holds alternative conceptions, but not fixed alternative frameworks, c.f. §2.4.3.) According to this view, young people's ideas are fragmentary, invented ad hoc, and have limited ranges of application (1993, pp. 46-47). Claxton suggests school children exhibit three categories of minitheories that he labels 'gut science', 'lay science', and 'school science' (p.50).
Claxton's gut science, which "stops you getting burnt and falling over" (pp.52-53), might be associated with Vygotsky's spontaneous concepts (§2.2.2), Pope and Denicolo's (1986) intuitive theories or Driver and Erickson's (1983) theories-in-action. It is "acquired through experience and is expressed in unreflective, unpremeditated action" (p.52, see also §2.4.4).
Claxton's lay science "gives you practical advice about when to plant the radishes, or how to load your camera; and … gives you intrinsically interesting things to talk about" (p.53, my emphasis). Claxton suggests that this type of knowledge does not have to be accurate, nor formalised, and it is not important "whether there is an inherent contradiction between what you are saying now and what you said yesterday" (p.53).
Claxton sees the role of lay science more in social terms as it "commonly comprises a store of 'amazing facts' that can be traded and discussed with others as a means of exploring or establishing friendships" (p.52), a point which has been taken further by Solomon. The notion of lay science has much in common with Solomon's notion of life–world knowledge (discussed below, §2.7.1) which she sees as "a rag-bag of odds and ends picked up from conversations of parents, teachers, and friends; from the television and magazines" from which children can select (1993b, p.88).
For Claxton (and as we will see below, for Solomon in relation to life-world knowledge), the criteria of formal science – that is "demands for rationality, logic, coherence, rigour, precision and explanation in terms of a limited set of agreed, technical concepts" – are of little importance in relation to gut science and lay science (Claxton, 1993, p.52). Collectively these two categories could be identified with children's science, in the sense that they compose those views of the natural world and the meanings for scientific words held by children before formal science teaching (Gilbert et al., 1982, p.627).
By contrast Claxton's third category, school science, "is articulated, consciously and deliberately transmitted and received" and Claxton suggests that if it is understood it forms "a highly coherent set of ideas" (1993, p.52.) Like Gilbert et al. (§2.3.2), Claxton is careful not to suggest that 'school science' is identical to scientists' science (pp.50-51).
Claxton's argument – that what the A.C.M. researchers have called children's science is in the form of many piecemeal, local, fragmentary minitheories – will be addressed below. However Claxton also asserts that 'school science' may be of this form. This does not seem to be consistent with his own statement that – when understood – 'school science' forms "a highly coherent set of ideas" (1983, p.52.) Perhaps Claxton means to imply that learners do not generally reach such an understanding, and so for typical school students their learning of school science does indeed make up just another cluster of minitheories.
However it is harder to accommodate Claxton's suggestion that 'scientist's science' should be considered as a 'fourth cluster' of mini-theories (one of which "both school children and school teachers are largely ignorant", p.50). This suggests a radical interpretation of Claxton's ideas, that all human knowledge is stored as discrete 'minitheories' – although presumably those categorised to be in the 'scientists' science' cluster are compatible enough to give rise to coherent and logical thought when accessed in the process of thinking.
Some of the characteristics of Claxton's minitheories, the lack of apparent coherence, and the absence of application of a basic model across a range of phenomena, have been recognised by Driver (Driver et al., 1985b, p.3). However she has also pointed out that "many notions children hold are used in a range of situations and have the characteristics of elementary models or theories" (Driver and Easley, 1978, p.62). Similarly, the Surrey group referred to "conceptual structures which provide a sensible and coherent understanding of the world from the child's point of view", and which could be held "very strongly" (Gilbert et al., 1982, p.623), so that "the person resolutely holds on to the original model and rejects those of others" (Pope and Gilbert, 1983, p.201). These workers do not necessarily deny that some aspects of children's thinking have the characteristics of Claxton's minitheories, but they suggest that not all of children's ideas in science can be dismissed so readily.
§2.6.1: A synthesis of Claxton and the A.C.M.
I will treat two aspects of Claxton's ideas separately. The distinction between his different clusters of minitheories reflects Solomon's ideas which are discussed and criticised below (§2.7, §2.8). However the notion of minitheories itself requires some comment here.
Both Claxton and those proposing alternative frameworks (Driver, Gilbert, Watts etc.) base their interpretations on empirical data. There is no contradiction if it is accepted that: the alternative conceptions that make up children's science may sometimes take the form of logically connected alternative conceptual frameworks, but may also take the form of discrete and isolated knowledge fragments.
This synthetic position may be understood in two ways:
- cognitive structure may be more integrated and coherent in some learners than others;
- in an individual learner different areas of knowledge may be represented in cognitive structure with different degrees of integration and coherence.
In the former case the overall degree of integration may be one aspect of intelligence (cf. Gould, 1992 {1981} , pp.234), and may relate to metacognitive skills (§2.3.10, and see also §2.8.5, §2.11.3), and may relate to levels of maturation (e.g. cf. Piaget's notions of development, and Vygotsky's notion of scientific concepts providing a framework for the learner to gain conscious access to spontaneous concepts, §2.2.1, §2.2.2). The empirical data that has been considered by Claxton, Driver and others has mostly been elicited from school children. However, it has derived from learners of very different ages, levels of maturity, interests and aptitudes in science.
The empirical data presented in this thesis concerns A level chemistry students, who as a group may be considered to be (a) above average intelligence, (b) relatively mature (16+ years), and (c) to have shown an interest in chemistry. These are types of the learners who might be most expected to hold alternative conceptual frameworks, even if much children's science is generally in the form of minitheories.
If the synthetic position is accepted, then a key issue for researcher in the field, is to identify the status of elicited ideas. As Driver and coworkers have suggested,
"one of the problems involved in investigating children's ideas is devising ways of probing thinking which enables [sic] us to sort out the status of the responses we obtain; to distinguish between those ideas which play a significant part in the thinking of an individual or a group and those which are generated in an ad hoc way in response to the social pressure to produce an answer in an interview or test situation"
Driver et al., 1985c, p.196
§2.7: Solomon’s criticisms of personal construction of knowledge
Solomon has been critical of the work of the 'alternative constructions movement', and has suggested that the notion of the pupil as scientist (Driver, 1983) is seriously flawed. Although Solomon's position is constructivist it is based in the social (i.e. inter-personal) rather than the (intra-)personal construction of knowledge. Her perspective therefore owes much more to Vygotskyan rather than Kellyan ideas (§2.2.2 c.f §2.2.4). By considering Solomon's ideas in some detail it will be possible to explore some of the central debates about the construction of knowledge.
§2.7.1: Solomon’s two distinct systems of knowledge
One of the outcomes of children's science interacting with formal instruction identified by Gilbert and coworkers (see above, §2.3.4) was the two outcomes perspective where pupils learn presented theories and explanations, and can use them in class and in tests, but revert to their existing ideas in everyday conversation and problem-solving (1982, p.624). Solomon has suggested that one should distinguish between what she labels the natural attitude and symbolic universes of knowledge. The natural attitude is characterised as to categorise experience loosely, to typify, and to absorb knowledge into fragmented meaning structures. In Solomon's scheme this leads to life-world knowledge which is reinforced by communication and language, and has persistence and social value. In contrast symbolic universes of knowledge (such as the theories of formal science) are fragile, have low social value, and an overarching nature. Solomon suggests that "when students learn the new formalism of scientific thought they store it in a different compartment from that of the familiar life-world thought of daily discourse" (1993b, p.96). She cites as evidence – a phenomenon she assumes familiar to teachers – learners suddenly being cued "into the domain of science knowledge" when "a whole network of meanings, theories and concepts are recollected and furnished with examples" (p.95).
According to Solomon, the domains of life-world and symbolic knowledge are dissimilar in genesis and mode of operation – and crossover involves discontinuity of thought. Solomon's comparison of these 'two worlds of knowledge' is presented in table 2.1.
| Solomon's comparison of knowledge | – in two domains |
| life-world knowledge | scientific knowledge |
| Social exchanges try to achieve a mutual understanding and agreement. | The aim of debate is to sharpen differences and to confirm or refute rival opinions. |
| Words used have multiple meanings which are not defined but negotiated socially. | Concept words are unambiguously defined for exact use. |
| Meanings are dependent on the cultural group and on the physical or affective context. | Concept meanings are symbolic and abstracted from any particular situation. |
| Apparent contradictions are tolerated. No logical method is thought to be needed. | A tight logical network of concepts and theories is claimed. |
| This knowledge system is well socialized by daily use with familiar people. | This knowledge is not well socialized since its methods are rarely used and then only by teachers outside the peer group. |
| Solomon, 1993b, pp.92-93. |
She suggests these domains of knowledge represent more than just discrete frameworks or separate clusters of minitheories, but two different systems of knowledge (1994, p.8). This reflects findings from mathematics education that suggest totally different strategies may be used to solve arithmetical problems in school and in 'street' contexts (Carraher et al., 1991).
From work with secondary pupils asked about energy changes, Solomon inferred that a lapse of time will select preferentially for the life-world structure of meaning if there is no further reinforcement of symbolic knowledge; and that successful crossing over and back from one domain to another will be more difficult than continuous operation in one domain, and is indicative of a deeper level ofunderstanding (Solomon, 1992, pp.110).
2.7.2: Situated Cognition
Solomon's distinction between life-world knowledge and symbolic universes of knowledge may be illuminated by the notion of situated cognition, a perspective
that people have different 'ways of seeing' that are appropriate in different contexts (Driver, 1989, p.486). This is an area which has been reviewed by Hennessy (1993), who reports that both experts and lay people apply thinking that is honed in a particular problem-context (p.29). According to this view the reversion to children's science (or life–world knowledge, or lay science) outside of formal learning contexts would be expected as specialist knowledge – such as formal scientific knowledge – is not considered relevant to everyday life, and does not tend to be activated in the absence of the – perceived – appropriate context (p.24).
Hennessy's review supports Solomon's emphasis on the difference in origins of life– world and symbolic universes of knowledge: the former having developed through solving problems in real-life contexts (Hennessy, 1993, p.30), whereas the latter would need to be "reconstructed" and re-contextualised before it could be used in everyday life situations (p.26). Indeed Hennessy describes schooling as "a unique culture, a specialised practice with its own conventions, organisation and concerns, which are in fact of little value to society outside" (p.2).
According to the situated cognition perspective the apparent 'partial, incoherent or internally inconsistent' nature of many alternative frameworks is to be expected as "pieces of knowledge or models are being drawn upon flexibly and according to their appropriateness and usefulness in a specific practical context" (Hennessy, 1993, pp.6-7, c.f. Claxton's ideas, above, §2.6). The situated cognition perspective therefore has an important message for those assessing learners (either as part of teaching, or researchers), as assessment activities are situated – that is the content and context of assessment activities have a strong effect on outcome (p.10).
It is worth pointing out here that, to borrow an aphorism, 'context is in the mind's eye of the beholder', for as Edwards and Mercer (1987) point out context is largely a mental phenomena that is not available to other people (pp.65-66). It is something that is "problematical" (pp.160-161).
In so far as Solomon's position is informed by notions of situated cognition she provides a way of explaining why the two perspectives outcome (§2.3.5) may be a consequence of teaching learners holding children's science. However, Solomon also emphasises the social nature of knowledge construction (c.f. §2.2.2), and argues that whereas discourse is generally concerned with finding consensus and common understanding, in science very different norms apply.
§2.7.3: The social construction of consensual knowledge
Solomon argues that "the process by which children construct notions for explaining the meaning of events in their daily life is more social than personal" (1993b, p.86). She points out that in the 'life-world' "it is taken for granted that others see things very much as we do", and we "expect to be able to understand each other and to share meanings" (p.86).
Solomon bases her argument on episodes from classroom discussion – collected during her own research – where she commonly observed an "unstated pressure" to resolve any disagreements (p.88), so that during a process of discussion contradictory opinions were often supported by the same children, with various assertions being put forward until some suggestion receives "social recognition". According to Solomon "familiarity wins the day" and unresolved disagreements were ignored (p.88).
As Solomon points out, this is not how scientific debate is meant to occur and the purpose of discourse is therefore different in the two domains, so that even a term such as 'explanation' takes on a different meaning – as "only in science … does 'to explain' mean to fit the event into a metaphorical scenario" (1992, p.107). According to Solomon scientific knowledge is by its very nature less likely to be the domain of knowledge called-upon by most people in most circumstances. Her argument is that whereas "life-world knowledge is 'learnt' through social reaffirmation in everyday situations, the more esoteric knowledge of science is the product of school learning – a later, secondary process of socialization" (1992, p.112).
From this social constructivist position, Solomon has criticised the personal constructivist (or as Solomon would say individual constructivist) approach of workers such as Driver, Gilbert and Watts. She has argued that Driver's Kellyan notion of the pupil as scientist is untenable: and she proposes "three troublesome questions" that need to be asked of the personal construction of knowledge position (Solomon, 1993b, pp.85-6):
1. If children's notions have been assembled in such a logical, almost scientific way, why do school children then have such difficulty in understanding the logical method of science and resist changing their notions in the light of new and compelling evidence?
2. If they have tested their ideas in the different circumstances of daily life why is it that they apply them so inconsistently?
3. If every child is his or her own independent scientist, how is it that within a cultural group notions are so much more similar than they are across different cultures?
For Solomon these questions are not answered by Driver's (1983) notion of pupil as scientist but rather it is "the sharp contrast" between the ways in which knowledge is constructed in the two domains which acts as a barrier to the learning of curriculum science (Solomon, 1992, pp.37-38). She argues that the self-contradiction found in learners' ideas should deny them the label of children's science (1994, p.9).
§2.8: A critique of Solomon’s position
Although Solomon's recent writings (e.g. her 1992, 1993a,b; 1994) criticise personal constructivism and establish an alternative position, it will be argued here that much of her work can be seen as compatible with the position taken by those researchers she criticises. There are some significant differences, and these will also be considered.
§2.8.1: The construction of scientific knowledge
Solomon has emphasised the distinction between life-world and scientific knowledge, and has suggested that scientific modes of thinking do not come naturally to the untrained. The recognition that scientific thinking is something other than 'common sense' has recently been the basis of a book by Wolpert (a professor of biology as applied to medicine), who suggests that scientific thinking and common sense are "not congruent" and points out that brains have been selected by evolution for survival in the natural environment, and that the ways of thinking that evolved did not (for most of humanity's existence) involve formal scientific thinking (1992, p.11).
However, the metaphor of pupil-as-scientist does not rest on pupils being accomplished scientific thinkers. Indeed Driver, in her The Pupil as Scientist?, describes the limitations of student thinking (1983, pp.61). More recently Driver and coworkers have pointed out that "learning science involves being initiated into scientific ways of knowing" (Driver et al., 1994b, p.3) – a statement that reflects Solomon's position – and have carried out an extensive research project into pupils' understandings of the processes of science (Driver et al., 1996).
The personal constructivist position does not depend on learners following a particular scientific heuristic, nor being able to apply some objectively rational and logical analysis in their modelling of the world. Rather it is sufficient that learners spontaneously try to make sense of their world by constructing internal models to help them "predict and control" (Kelly, 1963 {1955} , p.5) their surroundings, and then being able to evaluate and modify these models as required (Pope and Gilbert, 1983, pp.196-197). It was in this sense that Kelly suggested one should explore the notion of 'man-as-scientist', not that all people were in some sense perfect scientists. Indeed, as Pope has pointed out, few professional scientists can claim to have been so objective,
"For Kelly, the construction of reality is a subjective, personal, active, creative, rational and emotional affair; and if we are to believe modern philosophers of science, then similar adjectives can be applied to scientific theorizing and methodology."
Pope, 1982, p.6
Mahoney (1976) has suggested that many scientists use research methods which are "blatantly illogical" (p.153), something he attributes in part to the "logical fallibilities of individual scientists" (p.154). This author cites studies to demonstrate examples of such failings, and notes that as scientists are seldom given any formal training in logic, it is surprising that there is such a high expectation of the rationality of their work (p.154).
So the distinction between formal scientific thinking and 'life-world' thinking is neither so complete, nor so crucial as Solomon suggests. Indeed whereas Solomon wishes to emphasise the differences between children's learning and scientific thinking Strike and Posner put the opposite view,
"questions having to do with individual learning have certain generic structural features, whether they concern a scientist struggling with a new idea on the forefront of knowledge or with a child trying to understand elementary concepts about motion"
Strike and Posner, 1985, p.213
§2.8.2: The quest for common knowledge
Another key point in Solomon's position is that the purpose of communication in science is to "sharpen differences" rather than to "try to achieve a mutual understanding" (1993b, p.92). Her point is that in professional scientific discourse debate takes on a dialectic nature, that correspondents seek to take contrary views to test out positions. In normal social chat the purpose is quite different – to achieve a consensus, and preserve social cohesion.
However, although the two 'purposes' Solomon identifies are distinct, they do not directly map onto 'scientific' and 'life-world' exchanges. For one thing people can on occasion seek to avoid consensus in the life-world rather than reach it. Solomon acknowledged examples of this in the classes she observed (1993b, p.88) and suggested that this usually occurred where there was animosity between the children concerned (p.88). However, this proviso could also be applied to the debate between Newton and Leibniz, which although conducted on scientific and philosophical grounds, is recognised to have been less motivated by a desire to further science, than by personal animosity (e.g. Park, 1988, pp.221). This ill-feeling arose from a priority dispute, one of many such disputes in the history of science (Mahoney, 1976, p.119). The desire for consensus and social harmony is not always present in either the scientific or the life-world domain.
Solomon's distinction between the two domains also fails in another sense. Ziman recognised the same pressure for consensus in science that Solomon claims characterises the life-world. He sees the aim of science as "a consensus of rational opinion over the widest possible field" (Ziman, 1978, p.3, my emphasis), and describes scientific knowledge as being the product of a social and cooperative process,
"scientific knowledge is the product of a collective human enterprise to which scientists make individual contributions which are purified and extended by mutual criticism and intellectual co-operation"
Ziman, 1978, pp.2-3
That is, scientific dialogue is characterised neither by the quest for disagreement nor agreement per se, rather both are part of a process of constructing public knowledge. Solomon herself seems to accept this when she reports that "scientific epistemology now rests on … grounds which are social and consensual rather than objectively true" (1994, p.14, my emphasis), and opines that the solitary experiences of the knowing individual "will not do to describe either everyday knowledge, or scientific knowledge, or the learning of school science" (p.15, my emphasis).
Solomon's explanation of this discrepancy in her position seems to be that "science itself has been built up into a knowledge system by a consensual process which is not that of the life–world" (Solomon, 1994, p.16, my emphasis). Solomon acknowledges Ziman's work, and explains that science is a "corporate enterprise" and acknowledges that the scientific community has established means of monitoring the products of science (p.16). Yet it is difficult to see why this needs to be – in principle – a different process to that occurring in a discussion group set up in a school science class. Certainly the process would be different in degree, but not necessarily in type. Presumably the 'established ways of 'monitoring' the public constructions of scientific knowledge include observations of the natural world, and hypothesis formation and testing through experiment: the same ways that school pupils are being asked to monitor their own constructions in school science schemes developed from within the personal constructivist frame (Driver and Oldham, 1986).
Although the procedures of professional scientific activity do not match the social agenda of the school playground, nor do the activities that occur in a school classroom or laboratory. For example Edwards and Mercer note how school pupils recognise and accept that the world of the classroom has its own rules and agenda,
"the principle, to put it crudely, that lessons are about what happens in lessons appeared to be a ground-rule that the pupils had themselves acquired"
Edwards and Mercer, 1987, p.78
In contrast Solomon suggests that for some children only familiar with the construction of knowledge in the life-world the presentation in science classes of the "harshly uncompromising" process of working towards scientific knowledge "amounts to an affront to the social mores" (1993b, p.100). However the work of Edwards and Mercer suggests that there is no problem here as both teachers and pupils understand the 'game' through which teachers lead their pupils to 'discover' the accepted public knowledge that makes up the curriculum. Pupils' conversations in the playground may reflect Solomon's social agenda, but in the classroom pupils would be expected to recognise that they are being asked to talk to a different purpose (although they may not be very skilled in carrying out the tasks). It has been suggested to me that a useful distinction to make might be between knowledge which might be constructed in classrooms though the dialectic of teachers questions and pupils responses, and opinion which may be transient, and constructed more through social consensus.
Materials produced by Driver's CLiS project contain samples of dialogue from secondary pupils discussing the merits of their scientific ideas. These examples of pupil talk do include agreements, but also many cases of pupils putting their alternative opinions, and challenging the previous speaker. There are also many examples where one pupil is clearly asking for clarification or further exposition of another's ideas, rather than just looking to agree or disagree. My reading of the extracts in Brook and Driver 1986, Johnston and Driver 1991, and Wightman et al. 1986, certainly suggests that much pupil discussion in science classes can not be explained purely in terms of the social imperative of the life-world.
One might ask how Solomon could interpret her classroom observations so differently from Driver and others. Perhaps she notices and emphasises different aspects of classroom talk because she is approaching her research from a different perspective (1993a). Whereas she characterises the personal constructivists as following the ethnographic approach, i.e. "to ask children to explain their ideas and then listen carefully to their words in the verstehen tradition" (p.1), she herself is working in a 'cultural effects frame', that is "concerned with the children's ideas about science as reflections of the social influences and informal instruction which are at large within the community" (pp.7-8, my emphasis). For example Solomon discusses one particular example where a pupil, Mark, had changed his lone opinion towards a general consensus view (that energy was stored in food),
"Whether Mark had achieved understanding, or had merely moved in order to 'stand in with the crowd', is impossible to tell. Possibly the question does not even have real validity in the context of the social construction of knowledge where several concurrent meanings exist and local communication is a major objective."
Solomon, 1992, p.76, my emphasis
The question Solomon refers to in this extract would be of crucial significance to a researcher in the context of personal construction of knowledge, as such a researcher would wish to know whether the pupil had genuinely come to agree with his peers, or had just wished to avoid a public disagreement. One can also ask which perspective has more significance for how the pupils will perform in tests when they have to work alone: i.e., unless the socially constructed knowledge becomes personal knowledge it will not benefit children when they are assessed in formal examinations. However, from a perspective where meaningful knowledge construction is about reaching a common understanding Solomon is able to ignore this issue.
So then for Solomon, the solitary experiences of the knowing individual "wi$ not do to describe either everyday knowledge, or scientific knowledge, or the learning of school science" (Solomon, 1994, p.15, my emphasis). With respect to the research reported in this thesis, Solomon's perspective provides a useful warning not to over- interpret the significance of individual learners' utterances. However, for research that is based upon in-depth interrogation of individual learners, over extended periods of time, the ability of my colearners to present consistent, coherent and idiosyncratic pictures of their thinking – in the absence of a social group to reinforce a consensus – it is Solomon's emphasis on "consensus building" as "a process which might completely by-pass cognitive structures" (1992, p.75) which "will not do".
§2.8.3: Constructing knowledge in the classroom
At this point it is illuminating to consider how constructivist perspectives on learning might inform classroom practice, as this is another area where Solomon has been critical of the alternative conceptions movement. Solomon would argue that as scientific knowledge is "harshly uncompromising", the teacher's role is to direct the pupils to "make the imaginative but agreed pictures of consensual science their very own" (1995, pp.16-17) through questions "designed to elicit the right answer" (1992, p.132).
Edwards and Mercer point out that questions are components of "the discursive weaponry" that teachers use to direct classroom discussion towards the intended "common knowledge" (1987, p.46). They describe education as "essentially a process of cognitive socialization through language" whereby classroom discourse is used for "introducing pupils into the conceptual world of the teacher and, through her, of the educational community" (p.157). Part of the teacher's role is to assist pupils to take up as their own the desired vocabulary, and the selected descriptions and explanations that will form the "basis of joint understanding" (p.151).
This perspective may be compared with a naïve personal constructivist approach that given the right classroom experiences the learner will personally construct the desired knowledge. It is unlikely that this has ever been seriously proposed as an unproblematic process. The CLiS project was devised to "develop revised teaching approaches which would be informed by research on children's thinking in science and current theoretical developments in cognition" (Driver and Oldham, 1986, p. 105). These approaches would involve (p.108):
- devising learning materials which take account of students' prior ideas;
- developing ways of working in classrooms which encourage students both individually and collectively to become active in the learning process;
- making explicit the implications of adopting a constructivist perspective for classroom practice.
A basic tenet of this approach was that the curriculum should be a programme of activities which encourage pupils to (re)construct scientific knowledge (p.112-6). The teacher's role was to be a "facilitator" who would provide the appropriate opportunities for the pupils to undertake the construction (p.116). The constructivist model proposed included elicitation of ideas, and the restructuring of ideas – including exposure to conflict situations and construction and evaluation of new ideas (fig.3, p.119).
Solomon has pointed out that for a teacher to be aware of learners' ideas is not the same as having a means of bring out the desired changes (1994, p.10). Whilst acknowledging that CLiSP had produced "a rich source of valuable data" she still found it difficult to understand what was meant by 'constructivist teaching' (p.11).
Such an approach to curriculum planning and development is based on a view of learning as conceptual change (Driver and Oldham, 1986, p.117), a view that has been considered by a number of workers (see Gilbert and Pope, 1986a; Nussbaum and Novick, 1982; Strike and Posner, 1985). The approach discussed by these authors is to ensure students are aware of their ideas, then to challenge them with experiences that do not match their expectations (to produce what is variously called 'disequilibrium', 'cognitive dissonance', 'cognitive tension' or 'conceptual conflict') so that the learner will be motivated to test alternative models that do not conflict with experience.
Solomon has criticised constructivist teaching approaches on the basis that individual learners may have no stable alternative frameworks for a topic, yet the classroom activities intended as elicitation provide just the social context where the construction of alternatives to science will take place. Solomon also points out that interviews – so commonly used in constructivist research to elicit learners' ideas – are not part of the normal teaching repertoire (1994, p.10); but this is confuse research and teaching, and ignores the ability of interviews to remove the learner from the very social milieu that Solomon considers the trigger for learners' alternative ideas.
Even if Solomon were correct that the 'elicitation' process in constructivist teaching schemes is actually a 'construction' of alternative theories, this may not negate the approach, as Ault, Novak and Gowin had found that that the acquisition of a complex scientific concept seemed to be more likely where the learner had been able to produce a range of relevant ideas at an early stage (see §2.3.11). In particular, they found that it was not important if the learner's initial ideas were incorrect from a scientific viewpoint, as long as there was 'rich conceptualisation' on which to build,
"what matters most in the improvement of understanding is not simply the accuracy of conceptualisation, but the richness; not the sequence of acquiring new meanings, but the concerted effort to reconcile new with old; not the characterization of children's understanding chronologically, but the teaching of concepts by someone who takes time to find out how children modify meanings conveyed, how they apply concepts to make sense of events, and how they arrive at the claims they make"
Ault, Novak and Gowin, 1984, p.460, my emphasis
Although Ault and coworker's comments appear to negate Solomon's concerns, her doubts about 'constructivist teaching' is shared by Millar, who does believe in the individual's personal construction of knowledge. Like Solomon, Millar has pointed out that taking learners' ideas seriously needs to be reconciled with science as (at any one time) a body of knowledge that is to a great extent consensually accepted by the scientific community (1989, p.588). Millar's acceptance of the construction of knowledge as an intrapersonal process leads him to not automatically preclude particular teaching approaches. If the construction of knowledge is a mental process within an individual's mind, then a traditional method such as 'chalk and talk' might on occasion provide a suitable impetus to conceptual change,
"the process of eliciting, clarification and construction of new ideas takes place internally, within the learner's own head … independent of the form of instruction"
Millar, 1989, p.589
(It might be noted here that Ausubel (1961) has similarly pointed out that meaningful learning (§2.2.5) in adolescent learners does not rely on 'discovery' methods, and indeed he suggests that verbal reception learning can be the most efficient way of meaningfully learning subject content.)
Millar concludes that "science should be taught in whatever way is most likely to engage the active involvement of learners" (p.589). However he points out that all learning involves the learner in reconstructing knowledge internally (p.592), and – unlike Solomon – he recognises the value to this process of the type of activities recommended by Driver and Oldham,
"the classroom activities suggested by the constructivists for eliciting, clarifying and reconstructing ideas become immensely valuable for the teacher who is monitoring and managing this reconstruction process"
p.592
§2.8.4: Personal construction within a cultural context
The third of Solomon's questions to the personal constructivists was "if every child is his or her own independent scientist, how is it that within a cultural group notions are so much more similar than they are across different cultures?" (1993b, pp.85-6). Solomon points out that the nuances which words carry vary in different languages, and that this may be related to some of the alternative conceptions associated with such words in different cultures (1993b, pp.89-92) as ideas leave 'imprints in the language' (1992, p.168). The importance of words as tools of thought was emphasised by Vygotsky (see above, §2.2.2), and in-so-much as different languages use different words and word-meanings and therefore "cut up the world in different ways" (in Kuhn's expression, 1970 {1965} , p.268), it is quite reasonable to expect them to channel thinking differently. Indeed as Polanyi pointed out, different languages "sustain alternative conceptual frameworks" (1962 {1958} , p.112).
A learner's cognitive structure may be viewed as much a part of the learning environment as other factors. Hewson refers to Toulmin's notion of 'conceptual ecology' where concept formation is interpreted according to "the varied mental sets of individuals which are a function of their inte$ectual and physical environment" (1985, p.153, my emphasis). This idea is also used by Strike and Posner who suggest that "understanding entails finding a niche within a conceptual ecology" (1985, p.219). They list the following as features of such a conceptual ecology: anomalies, analogies and metaphors, exemplars and images, past experience, explanatory ideals, general views about the character of knowledge, metaphysical beliefs about science, metaphysical concepts of science, knowledge in other fields and competing conceptions (pp.216-217). It will be seen that language has a role to play in many of these features.
In my view Solomon's question of the personal constructivists is misguided as it appears to imply that a commitment to personal construction of knowledge assumes the learner-as-scientist will be able to develop theories of the world based on experience of the physical world unmediated by cultural factors. This has never been the assumption: and Driver points out the inadequacy of such Baconian naïveté as early as page 4 of her "The Pupil as Scientist?" (1983). Solomon questions the child-as-independent scientist, whereas Driver clearly wrote about the pupil-as- scientist working within a community of scientists: "science as a cooperative exercise as opposed to an individual venture" (p.4, my emphasis). Although Hewson was writing some years before Solomon posed her question, the following quotation might stand as a response to Solomon's inquiry,
"The intellectual environment in which a person lives (including cultural beliefs, language, accepted theories, as well as observed facts and events) favors the development of some concepts and inhibits the development of others. Thus the intellectual environment acts as an ecological niche."
Hewson, 1985, p.154
§2.8.5: The value of Solomon's critique – learning science as cognitive apprenticeship
Although it has been argued that Solomon's perspective does not undermine the basic tenets of the (personal) constructivists, her emphasis on the role of social interaction may have been influential in emphasising this aspect of the individual's learning environment. Certainly writers such as Edwards and Mercer, and Hennessy, have put emphasis on the role of the teacher in providing the structure – in a Vygotskyan sense, that is "through a series of processes such as modelling, coaching, scaffolding, fading, articulation" (Hennessy, 1993, p.11) – to encourage the desired construction of knowledge. So, as Hennessy says,
"expertise is acquired through both the spontaneous invention of personal, highly efficient procedures in response to the needs of a situation, and through apprenticeship"
1993, p.15, my emphasis
Hennessy explains this notion of apprenticeship as involving "providing help in developing an appropriate notation and conceptual framework for a new or complex domain and allowing the learner to explore that domain extensively, then gradually withdrawing support" (p.12). Through this process the learner will develop "tacit strategic knowledge" both cognitive and metacognitive. This will include strategies for exploring new domains of knowledge, and for 'reconfiguring' knowledge in a topic area (p.20, my emphasis).
The position of Driver and her coworkers has developed somewhat to take such perspectives into account, so that the teacher's role not only includes providing physical experiences and encouraging reflection, but giving learners access to what they have called the "symbolic realities" or "cultural tools" of science" (Driver et al., 1994c, pp.4-5, my emphasis). This evolved constructivist perspective means that teachers should not just expect pupils to demonstrate alternative conceptions, but possibly multiple conceptual frameworks which are appropriate to different contexts (p.5).
The social constructivist perspective has been emphasised in the new presentation of the personal constructivist position, so that learning is seen to involve induction into the 'symbolic world' of science (p.5) as well as the social interaction in the classroom setting (p.6). Constructivist teaching still involves "challenging learners' prior ideas through discrepant events" but is also recognised as socialisation (c.f. Edwards and Mercer, 1987, see §2.8.2),
"young people entering into a different way of thinking about and explaining the natural world; becoming socialised to a greater or lesser extent into the practices of the scientific community with its particular purposes, ways of seeing and ways of supporting its knowledge claims"
Driver et al., 1994c, p.8
§2.9: Degrees of integration in cognitive structure
From the literature reviewed it is possible to develop a position which gives heed to the various perspectives discussed, and which does not unnecessarily limit the interpretation of the data collected as part of this research.
It is considered here that people can construct within cognitive structure extensive, largely coherent and consistent, frameworks of ideas (and there will be evidence for this presented in chapters 7 onwards). These frameworks may be analysed in terms of components at the level of conceptions – but the various conceptions are related through the framework.
However, it is not claimed here that a$ knowledge held by a$ people is structured in this way. As Claxton suggests (§2.6), knowledge may be stored as substantia$y discrete parts, as in his mini-theories, thus explaining the characteristics of research data that do not support the presence of 'frameworks'.
In other words, an overview of the research literature would lead to an assumption that there are different possible degrees of integration of knowledge held in cognitive structure. An 'ideal', totally integrated, representation of an individual's knowledge would have all its components inter-related though a logical, self-consistent set of links. At the other extreme would be the storage of knowledge as a random catalogue of totally independent fragments – as in Claxton's "store of amazing facts". Real cognitive structures would be between these extremes, and the degree of integration of knowledge in a subject area should be a major concern for teachers and researchers (c.f. §1.5.2). The integration of some new 'piece of knowledge' into the structure would depend upon the level of integration and coherence of the existing material, and the perceived relevance of the new information to existing
knowledge (§2.2.5). Degree of integration is one factor affecting the likelihood of learning impediments occurring during instruction (§1.5.2). The degree of integration of my colearners' ideas related to chemical bonding is one of the factors investigated in this research (see the case studies in chapters 7 and 8).
§2.9.1: Interpretations of multiple frameworks
Accepting the possibility of degrees of integration within cognitive structure, then there are four possible interpretations of multiple !ameworks (§2.5.2). If a researcher's data su&ests that a learner holds two or more frameworks to explain the same phenomenon, then this could be because:
- the learner's mental representation of the concept is unified and consistent, and consequently 'multiple frameworks' demonstrate inadequacies in the researcher's models (c.f. Pope and Denicolo 1986, see §2.5.2);
- the learner may not represent his or her knowledge of the topic in coherent and self-consistent terms, but rather as a range of discrete knowledge fragments: the researcher over-interprets a learner's utterances as a set of conceptual frameworks due to his or her assumptions about cognitive structure (c.f. Claxton, 1983, see §2.6);
- multiple frameworks reflect a genuine aspect of the learner's cognitive structure: learners can have alternative versions of 'the same' concept – and this may be seen as due to a lack of integration of concepts, and an immature stage in conceptual development (c.f. §2.6.1);
- multiple frameworks reflect a genuine aspect of the learner's cognitive structure: learners can have alternative versions of 'the same' concept – and this may be seen as an appropriate adaption by the learner to the different contexts in which the ideas may have to be applied (see §2.7.2).
All four of these possibilities are feasible, and they should not be seen as mutually exclusive positions. It is possible all four options occur (and may quite likely be represented in the literature). An awareness of these possibilities has informed the analysis of data collected in this present study.
Where the multiple frameworks are authentic, they may represent an appropriate level of understanding in view of the nature of chemical knowledge (see chapter 1, §1.3.1, §1.7.1). However, they may also provide indicators by which to measure progression, especially where the frameworks do not match the curriculum science models (see §8.4.3).
§2.9.2: Shifting between alternative frameworks
The possibility of learners' holding several frameworks for explaining a phenomenon does not rely on the learner being aware of the context-dependence of knowledge – it is not suggested that the learner necessarily consciously makes a decision to apply a life-world or an academic framework for thinking about, say, force. Indeed it would seem more likely that this is a tacit process influenced by a whole range of factors (who posed the question, how formal was the language used, in what location and situation the question was asked, etc.) The key point is that the learner has distinct alternative frameworks for thinking about what – to the researcher – is a single concept area.
Yet it is clear that such shifts between different frameworks of thinking do occur. For example the historian comparing how an event would be interpreted in the conceptual schemes ascribed to Aristotle, Galileo and Newton would need to be able to shift between the schemes (each of which must be represented in his or her own cognitive structure). The educational researcher needs to be able to shift into the model of the learner's thinking to try to understand what the learner means by his or her utterances. Kelly referred to subsuming the constructs of the other into one's own system (1963 {1955} , p.174), and this implies some sort of hierarchical arrangement of constructs, so that one may shift between frameworks by moving to a higher level of the construct system – perhaps what Pope and Denicolo refer to as "the system of necessary interrelationships" of "component intuitive theories" (1986, p.158). I have represented this diagrammatically in figure 2.1.
In this diagram the areas labelled as 2 and 3 represent some parts of cognitive structure assumed to represent knowledge in self-consistent ways – i.e. conceptual frameworks. However the knowledge represented in 2 need not be consistent with that represent in 3, because transfer between these regions is mediated by a link at a higher (more abstract perhaps in Kelly's terms) level in cognitive structure.
This model for the process might be seen as analogous to the idea of a metalanguage which allows one to discuss language. Although thinking within one region of the cognitive structure is constrained (channelled) by the logic of the constructs in that region, it is always possible to move outside that logic via a more abstract level. Without some such mechanism there would be no imaginative leaps, which are essential for the act of discovery – even at such a mundane level as understanding the punch-line of a joke (Koestler, 1978.)
§2.10: Conceptual change
As the true nature of cognitive structure remains unknown, it is not possible to know exactly how conceptual development can occur (and models such as that presented in figure 2.1 above and fig. 2.2 below are conjectural, although they may have considerable heuristic value in guiding research). For one thing, there is no precise agreement on what a concept is (see §2.10.2 and §4.7.2), and therefore how it should be understood to be represented in cognitive structure.
§2.10.1: Two types of conceptual change
Conceptual change may be considered to be of two types. Firstly, a great deal of learning may be understood as 'local', in that it concerns a small addition to, or alteration of, knowledge, which – to a first approximation – has no repercussions for overall conceptual structures. Secondly there is learning which has greater ramifications: changes in perspective that fundamentally alter the perceived relationship between different concepts held in cognitive structure, and suggest that some form of major reorganisation is required (e.g. Novak, 1985, p.193-194). For example, Ault, Novak and Gowin found that sometimes "acquisition of a key concept causes a significant shi% at a number of levels in the organisation of conceptual understanding" (1984, p.459, my emphasis).
Probably much learning is actually intermediate in nature (Strike and Posner, 1985, p.216), but these two classes of change, akin to accretion of knowledge and conceptual revolution, suffice to stand for the spectrum of conceptual changes learners undergo. Although different workers have chosen to give various labels to these two categories, (for example, Ault et al., refer to progressive differentiation and integrative reconciliation, 1984, p.460) they may be seen as derived from Piaget's distinction between assimilation and accommodation – that is "incorporation of new objects and experiences into existing schemas" and "modification of schemas as a result of new experiences" (Beard, 1969, p. ix) respectively. The former type of change can readily be modelled, but the latter, is more problematic. Thagard's model of conceptual change is discussed below (§2.10.6), and his approach to modelling these two types of conceptual change is described in appendix 9 (§A9.2).
§2.10.2: The relational view of concepts
Gilbert and Watts criticised what they labelled the 'classical' view of 'concept' – that is "that all instances of a concept share common properties and that these properties are necessary and sufficient to define the concept" – as a gross oversimplification based on an assumption that knowledge is arranged
hierarchically in discrete layers in the mind (1983, p.65). They identified this approach as – in Kelly's terms – 'accumulative fragmentalist' (p.65), and suggested it leads to a research programme of spotting bugs in the system, that is 'misconceptions' (p.66, c.f. §2.3.1). They preferred an alternative 'actional' view of concepts (p.66, c.f. the 'theories-in-action' of Driver and Erickson, 1983, §2.6). They also acknowledged an intermediate 'relational' view, where conceptual structure took the form of a network (p,68), which enabled concepts, to have borderline cases (p.67). (In the present research one might give the example of the extent to which van der Waals' forces are included in the concept of the chemical bond.) However, Gilbert and Watts did not feel that the relational model could explain more radical reorganisations of knowledge (i.e. accommodation, §2.10.1).
One might suggest that a learner's developing concept of the chemical bond could largely be modelled from the relational view, that is, as,
"a person's experiences accumulate, a concept changes in the number of exemplars and their degree of membership but remains essentially the same concept in terms of its internal features and its external links."
Gilbert and Watts, 1983, p.68
If cognitive structure is viewed as a network, then the meaning of a concept depends upon the whole network of propositions that it is part of, so that if any connection is altered, the meanings of all the interrelated concepts shift to some extent (Phillips, 1987, p.206). This perspective is referred to as 'semantic holism'. The development of the chemical bond concept to include hydrogen bonds might be conceptualised in these terms. The learner's cognitive structure may be understood to include a network of propositions relating to chemical bonding. The assimilation of the hydrogen bond concept into the subsuming chemical bond concept (by adding a new proposition – hydrogen bond is a type of chemical bond – to the network) would change the meaning of 'chemical bond' to the learner. To the extent that the concept 'covalent bond' is connected in this network its meanings would also shift (though in this case the change in meaning is more subtle). Gilbert and Watt's criticism was that such a model would not seem sufficient to explain more radical reorganisations of knowledge (p.68): an example might be how a learner might accommodate a new theory of bonding based on molecular orbital theory, or how a learner might switch from seeing covalent-ionic as a dichotomy to a continuum (see appendix 4 for an analysis of how the bonding concept might be developed during an A level course).
§2.10.3: Models of conceptual change
A number of workers have tried to explain conceptual change as a rational process, based on decisions about the relative merits of different conceptions (§2.10.5). Gilbert and Watts (1983) considered several approaches to conceptual change. One perspective is analogous to views of evolutionary (i.e. Darwinian) development, and had been applied (by Toulmin) to historical development of scientific concepts. In this model there is a continual generation of conceptions, some of which are selected for (as they show greatest value in problem solving) and retained, whilst most are discarded (Gilbert and Watts, 1983, p.89). Again such a perspective could not explain the radical conceptual changes of accommodation (p.68).
Gilbert and Watts discussed a 'catastrophe theory' model of conceptual change, which could explain the more 'revolutionary' changes in a learner's thinking (an approach since discussed by Boyes, 1988). This is basically a cost-benefit model: an existing notion is held until the point is reached where the benefits of change outweigh the cost (Gilbert and Watts, 1983, p.91). The catastrophe theory model was seen by Gilbert and Watts as having the most potential as it could explain both gradual and sudden shifts in ideas. A possible difficulty with such a perspective is that an alternative view is only accepted if it is seen to be advantageous, but until it has been built up and explored it is unlikely to be judged as a serious competitor for a view that is already held and grounded in one's experience. This objection can be overcome if one distinguishes between the construction of a perspective, and the acceptance and belief in that view (Driver and Bell, 1986, p.451), and if one accepts that people readily construct inconsistent notions into their cognitive structures.
If a learner's entire conceptual structure was required to be to be unified, consistent, coherent etc., this might lead to a very inflexible approach to learning. Once a conception had been acquired and represented in cognitive structure, it would not be possible to accept any contrary conception without discarding the original. Any two conceptions which are inconsistent in terms of the whole system of logical relationships would not be tolerated in structure at the same time. This would require a highly effective 'logic checker' monitoring the entire knowledge structure, and some sort of 'decision unit' that could decide which conception to jettison. When new information apparently contradicts existing knowledge, an immediate decision, to ignore the new idea or to discard an existing conception, would be needed for an individual's mental representation of a concept to remain unified and consistent. In view of the imperfections in our ways of coming to knowledge such a cognitive system would be highly inflexible: in a world where the information available is often incomplete or seems contradictory it would also be highly inefficient.
Alternative approaches suggest how learning may take place in such a 'fuzzy' information environment. If inconsistency is tolerated it is possible to consider alternative possibilities until there is judged to be sufficient evidence to make an informed choice between them. In the meantime the consequences of the alternatives can be explored. (This raises interesting questions as to the extent to which such processes are subconscious, systematic, intermittent, prioritised by current concerns, etc.)
A model of cognitive structure which enables knowledge to be stored in discrete fragments allows us to understand how we may learn about two apparently separate phenomena, and then later come to know that the two are connected, and are aspects of some superordinate concept. The lack of coherence in learner's ideas that has been commented upon by some researchers (such as Claxton, §2.6), can be understood as separate storing of ideas as discrete conceptions, or as parts of distinct conceptual frameworks. This has particular significance for the present research, as for example when covalent and ionic bonding come to be understood as extremes of a single model (rather than explained by distinct models) or when intermolecular bonding and atomic binding are understood to be due to the same basic physical forces as intramolecular bonding. At the point where the two fragments become related in cognitive structure it is possible that inconsistencies and contradictions will come into being. This is illustrated schematically in figure 2.2.
Part (a) represents two discrete conceptual frameworks ➀ and ➁, each comprising of a set of internally consistent conceptions. In (b) the learner has acquired a new conception which links ➀ and ➁, and which enables him or her to start to perceive them as aspects of the same concept. At this point there may be inconsistencies in the new conceptual framework as parts of ➀ and ➁ are not logically compatible. Part (c) reflects a later time when the learner has changed some of the propositions that were components of ➀ and ➁, so that they are consistent; and has explored the new unified concept area ➂, and discovered further connections that were not apparent when ➀ and ➁ were perceived as unrelated.
As an example, ➀ might represent a learner's ideas about gravity, and ➁ her ideas about orbital motion. She might be told, and accept, that satellites remain in orbit due to gravity, and thus a link is made between these two frameworks of knowledge that will in time lead to an enlarged framework of knowledge about gravitation (➂) which incorporates most of her old ideas about gravity and satellites that were previously considered unrelated. In the model, when the link is first made, and before there has been any opportunity for other changes in ➀ and ➁, there are inconsistencies in the combined framework. Perhaps ➀ includes a conception that gravity makes everything fall to the ground, and ➁ that objects stay in orbit forever. Clearly one or other or both of these conceptions must be altered before the new framework can be considered logically coherent.
Note that in this model there are times when the learner holds multiple 'partial' frameworks for a concept area (fig.2.2a), and during the process of integration inconsistent propositions are part of the same framework (fig.2.2b).
§2.10.4: Modelling learning impediments.
A similar type of diagram may be used to model the various aspects of a learner's cognitive structure that may act as learning impediments (§1.5).
Figure 2.3: a model of learning and learning impediments
(a) effective learning; (b) fragmentation learning impediment; (c) deficiency learning impediment; (d) ontological learning impediment; (e) epistemological learning impediment.
Figure 2.3 models an example of effective learning in comparison to the effect of the four classes of learning impediment.In this diagram the square at the left of the figure represents a small part of a learner's conceptual structure (shown as a discrete fragment for the sake of simplicity). In part (a) it is assumed that the learner's prior knowledge matches the desired prerequisite learning for the topic.The symbolism shown as being added (i.e. "+") in the figure represents the conceptual structure of the related segment of curriculum science that the teacher intends to teach.In (a) the desired learning takes place, and the square at the right of the diagram represents the 'target' conceptual structure with the new knowledge extending the existing prerequisite knowledge (which is shown otherwise unchanged, for the sake of clarity).In (b) the learner has the required prerequisite knowledge structure, but does not perceive the teacher's presentation as relevant to any previous knowledge. A fragmentation learning impediment means that at best the new knowledge will be learnt as an isolated fragment.In (c) the learner does not already hold all of the required prerequisite knowledge needed to make sense of the teacher's presentation in conceptual structure. A deficiency learning impediment means that the intended links can not be forged.In (d) the learner holds the intended prerequisite knowledge, but already has alternative conceptions relating this to the new topic area due to intuitive theories. An ontological learning impediment interferes with the intended learning.In (e) the learner holds prerequisite learning, but due to misinterpreting previous teaching already has alternative conceptions relating this to the new topic area. An epistemological learning impediment interferes with the intended learning.The situations shown as (d) and (e), the substantive learning impediments, are very similar (I have represented the former with additional links, and the latter with extended links, but this is somewhat arbitrary). It is possible that sometimes in this situation the learner may replace the inappropriate conceptions and form the target structure. However on other occasions the learning impediment may be only partially overcome, or may lead to fragmented learning, or – as in the diagram – no learning (c.f. 2.3.4).
§2.10.5: Rational criteria for conceptual change
In the model represented, the two distinct areas of cognitive structure in figure 2.2 were referred to as multiple 'partial' frameworks, as initially they were perceived as relating to separate concepts by the learner, but subsequently they reorganised into a single framework. It is possible to consider the situation where the two distinct frameworks were not unified (in the example, the learner did not come to see orbital motion as a gravitational phenomenon), but each developed until they both described and explained much the same range of phenomena. In such a case the two frameworks might come to have many similar elements, as well as aspects that were inconsistent. The learner would have available multiple frameworks for interpreting those phenomena within the range of both frameworks. (Perhaps sometimes one framework is based in everyday experience, or 'lay' science terms, and the other based on formal instruction, as discussed above, §2.3.5, §2.7.1, §2.7.2).
One could conjecture that whether partial frameworks are subsumed into an integrated structure, or developed extensively in parallel, is likely to depend on the point at which there is 'recognition' (by which I do not necessarily mean consciously) that the two frameworks are closely related: if this occurs before there is a good deal of redundancy between the frameworks the 'benefit' of the few changes needed for integration may outweigh the 'cost'. If the two structures are extensive when the realisation occurs, the disruption and effort of a major restructuring may not be justified. In this situation it is possible that multiple frameworks will be retained indefinitely, perhaps each accessed according to different contextual cues (c.f. §2.7.2).
Ausubel's meaningful learning theory (§2.2.5) and Kelly's theory of personal constructs (§2.2.4) would suggest that individuals are – in principle at least – motivated to make sense of their worlds. However, presumably, there is some 'effort' involved in undergoing conceptual change. This is certainly true in a physical sense (in terms of energy and entropy considerations). In terms of Kelly's P.C.T. (see above) the ease with which conceptual change may occur depends upon a feature of a person's constructs referred to as 'permeability', so that "the variation in a person's construct system is limited by the permeability of the constructs within whose range of convenience the variants lie' (Modulation Coro$ary, Kelly, 1963 (1955), p.77). According to Kelly's scheme, a construct is permeable "if it will admit to its range of convenience new elements which are not yet construed within its framework", and there are "relative degrees of permeability and impermeability" (1963 (1955) p. 79).
Strike and Posner (1985) have suggested that learning should be considered "a rational enterprise", where rationality is concerned with the conditions that should lead someone to change his or her mind (p.211). These conditions involve judging how well competing conceptions match empirical evidence, can explain experience, meet metaphysical assumptions about the form explanations should take, and are consistent with other knowledge (pp.212-215).
Strike and Posner suggest four conditions that must be satisfied before accommodation will occur (p.216; see also Thorley and Stofflett, 1996, for a discussion of these factors). Firstly the learner must have reason to be dissatisfied with existing conceptual schemes. They point out that accommodation is unlikely if existing frameworks can be made to work with minor adjustments (p.216). Secondly the learner must have 'minimal' understanding of the new conceptions, so that it's potential for explanation may be explored. They suggest that this involves being able to relate the new conceptions to some existing part of cognitive structure, and to familiar examples from experience (pp.216-219). Their third criterion was that the new scheme should seem a plausible alternative because it can be seen to explain the apparent discrepancies in the present scheme, and it meets metaphysical expectations. Finally the new conceptions should seem to be 'fruitful', in the sense that they suggests the possibility of wider explanatory scope (p.216).
§2.10.6: Explanatory coherence: an example of a specific model of conceptual change
In the literature reviewed so far the discussion of conceptual change has dealt with general principles. Thagard (1992) has produced a model based on similar principles to Strike and Posner, using the criterion of 'explanatory coherence' to determine when conceptual change would be expected.
Thagard's approach was primarily developed to analyse historical examples of conceptual change (see appendix 9, §A9.3) – to "understand the structure and growth of scientific knowledge" (p.3) – however he considers his approach to apply to contemporary learners of science as well as scientists of historical standing. His particular computer model assumes knowledge is arranged hierarchically – which may not always be the case (§2.10.2), but it is a useful example of how the general principles discussed above may be built into a functioning model that replicates conceptual change. (More detail of Thagard's model is given in appendix 9).
Thagard sees the personal construction of models of alternative scientific theories as a step in a rational process of paradigm shifts. The scientist – or young learner – holds one theory, but gradually builds up an understanding of, and familiarity, with an alternative. If the alternative comes to be seen as having greater explanatory coherence then it will become the preferred theory with which to operate in that domain. (Thagard's criteria for explanatory coherence are discussed in appendix 9, §A9.5.) Thagard describes how a scientist exposed to an alternative theory to the one held will construct a model of the theory 'in the background' to compare with his or her original (p.60).
For example when chemists learnt enough about the oxygen theory to believe it had greater explanatory coherence than the phlogiston theory, they changed to the new theory. For this to happen they had to be instructed in the new theory, but also had to have time to construct and explore, or read about and reflect on, the arguments in favour of the two alternative theories: "setting up the requisite nodes and links, was not enough: people had to use the new system enough to appreciate its power" (p.59). Thagard considers this to be a process which may take years. Priestley's rejection of the oxygen theory may be considered rational if it is understood that as the "preeminent phlogiston theorist" he had developed over many years the most elaborate and coherent conceptual scheme based around the phlogiston concept and therefore he never explored the oxygen concept enough to appreciate that it had greater potential (pp.59-60).
Thagard suggests similar processes may be operating in children, and conjectures that when they learn enough about an aspect of curriculum science to "consciously or unconsciously" appreciate it has greater explanatory coherence than their children's science they will switch to using the taught version (p.258).
Like Novak, and Strike and Posner, Thagard distinguishes two types of conceptual change (his taxonomy of epistemic change is reproduced in appendix 9, §A9.4). He considers adding or substituting a single concept or rule as relatively trivial, whereas 'revolutionary' changes which involve the overthrow of whole systems of concepts are more difficult to understand (p.6). Thagard models cognitive structure in the form of a network of concepts ("mental structures representing what words represent") connected by propositions ("mental structures representing what sentences represent", p.21), with the concepts making up the nodes of the network (p.30). The networks are primarily structured via kind-hierarchies and part- hierarchies" (p.7, see appendix 9, §A9.2). In such a model conceptual change is easily represented as adding or removing nodes and links (p.32), although these changes may be more or less severe depending at what level in hierarchy the change is made (p.34).
Thagard's analysis of historical case studies suggests a range of criteria are used to determine the explanatory coherence of a hypothesis" (p.63), and that alternative explanations 'compete' on such dimensions as,
• How much does the hypothesis explain?
• Are its explanations economical?
• Is the hypothesis similar to ones that explain similar phenomena? • Is there an explanation of why the hypothesis might be true?
(It will be noted that these factors are similar to those identified by Strike and Posner, see §2.10.4). The first of these criteria – "the explanatory breadth of the new theory" – seemed to be the most important factor (p.248). However, greater familiarity with the existing theory and its potential applications may act as a barrier (c.f. substantive learning impediments, §1.5.3). Thagard points out that it takes time and mental effort to explore the new ideas. This exploration may include debate with peers, and reflection on the discussion (1992, p.59, c.f. the discussion above of the validity of Solomon's claims about the nature of classroom discourse in science lessons, §2.8.3).
An important aspect of Thagard's model is his acknowledgement that during major conceptual change "the new conceptual system does not arise by piecemeal modification of the old one" but "rather, the new one must be built up largely on its own, and its replacement of the old is the result of a a global judgment of explanatory coherence" (Thagard, 1992, p.60, my emphasis). Thagard's model thus explains the epistemology of conceptual revolutions in terms of the construction of representations of alternative theories in a conceptual network (see appendix 9, §A9.6, c.f. discussion of 'multiple frameworks' above, §2.9).
Rowell and Dawson (1985) have suggested an approach to bringing about conceptual change in the classroom based on similar ideas. They suggest that once learners' basic ideas about a topic are elicited, they should be used to build the appropriate curriculum science model (scaffolded by the teacher). Once the learners have constructed the new model they are given the opportunity to practice applying it. Then (once it is familiar, and considered their own construction) the class are asked to compare the new model with specific existing conceptions, and again the teacher structures the discussion to bring out the advantages of the curriculum science idea.
§2.11: The assumptions made in the present research.
The literature reviewed in this chapter provides a theoretical base for the assumptions built into the present research. In view of the disparate views about the nature of learners' alternative ideas in science (§2.3.1, §2.3.9, §2.4, §2.5, §2.6, §2.7), and in view of the principles of 'grounded theory' work (§4.4) I have attempted to keep those assumptions as open-ended as possible so that my data analysis was not heavily handicapped by my own preconceptions. However, just like my colearners, I may not be fully aware of the biases built into my own conceptual system. I was personally introduced to many of the issues reviewed in this chapter during my first year of teaching when I attended a conference on Concept Development in the Learning of Physics. When I set out on the present research project, some years later, I remembered having been enthused by the conference, but did not consciously recall that I had been introduced to ideas such as minitheories (Claxton, 1983), Vygotsky's notion of language as tools for constructing knowledge (Sutton, 1983), alternative frameworks (Engel and Brook, 1983; Watts, 1983c), concepts carrying metaphysical historical baggage (Roche, 1983, c.f. Bachelard's epistemological obstacles, §1.6) and the distinction between life-world and scientific knowledge (Solomon, 1983).
§2.11.1: Theoretical position taken in the research
My position is constructivist, in the sense explained at the beginning of this chapter (§2.1.2), and in general derives from the work of the A.C.M. However, it is also a synthetic position that takes into account the criticisms of workers such as Claxton and Solomon, so that I do not assume that each of my colearners' utterances reflect stable and extensive individual alternative frameworks. Rather, the literature reviewed informs the present research in the following way:
- a learner's knowledge relating to chemical bonding could be stored in cognitive structure as a series of discrete knowledge fragments, which are perceived as having little if any relation to one another;
- development towards an 'ideal' understanding of this concept area would involve the integration of these discrete knowledge fragments into a single coherent and self-consistent model of 'chemical bonding'.
- to the extent that the topic is highly complex, and the information available to a learner at any level is incomplete and imprecise, (and the possibility that restructuring cognitive structure may be a long- term process) it is likely that total integration of knowledge will not be achieved.
- in the absence of full and reliable information, and in the limited time available to students on an A level course, the optimum model of 'chemical bonding' in a learner's cognitive structure may well be somewhat fragmented, and separate fragments could be incompatible to some degree.
Therefore in my research into learners' understanding of chemical bonding I will look to interpret the data in such a way as to admit the following possibilities:
- discrete knowledge fragments, of the form of isolated conceptions; and
- concurrent separate (i.e. multiple) conceptual frameworks;
and - an integrated conceptual framework for understanding bonding.
That is, I will not presuppose the extent to which my colearners' understanding of chemical bonding should be modelled as an inventory of unrelated conceptions, or as one of more coherent conceptual frameworks. The degree of integration of knowledge is a potential indicator of the development of understanding, but this must be interpreted in terms of the information available (at the level being studied) to learners about the topic. The extent to which the development of a learner's understanding can be interpreted as increased integration of knowledge is an empirical question for this present research.
§2.11.2: Working terms used to discuss research results
In view of the lack of consistent terminology in the field (see §2.4), I set out here the manner in which I will use terms. Following the principles outlined in chapter 1 (§1.4.1), in this study I assume that my colearners' knowledge was organised in cognitive structures that are not directly observable. In chapter 6-12 I will therefore not be discussing these structures themselves, but my models which represent my inferences about aspects of learners' cognitive structures.
Particular propositions made by colearners will be represented by the term 'conceptions' (with the proviso that the conception presented is my interpretation, not an element of cognitive structure itself). Where a range of propositions appear to be logically based on a closely related set of propositions, I will refer to these key propositions as forming an explanatory principle, and the larger network of related propositions will be called a complex. In the final chapter, chapter 12, I will consider whether the explanatory principles and complexes presented could appropriately be described in terms such as Gestalts or conceptual frameworks.
The stability and degree of integration of my colearners' thinking about chemical bonding, and the insight this may provide into any underlying cognitive structures, are empirical issues which will be explored through the research described in this thesis. The research evidence rallied to consider these issues in chapters 7 through 11 will inform the advice offered to teachers of chemistry in chapter 12.
§2.11.3: The colearner in the context of the research
Figure 2.4 summarises some aspects of my perspective on the present research which derives from the material presented in chapters 1 and 2.
This figure is intended to represent some of the major components of the context of the present research (and has some similarity to Osborne and Wittrock's 1983 schematic representation of their generative learning model). The central feature is the learner, the colearner in my study. In the research I observe the colearner's behaviour (5) and interpret this to develop my models of the learner's understanding. To do this I provide a structure – such as a set of interview questions about foci diagrams – as a context for the learner's thinking (1). As teacher- researcher I am fluid part of system as 1 and ➄ make up an ongoing discourse, and 1 is an interactive response to 5 as much as vice versa. In this way I attempt to probe the learner's Z.P.D. (zone of proximal development, §2.2.2). My questions and foci are mediated (2) through the colearner's sensory apparatus (e.g. his eyes and ears, and those parts of the brain which filter and interpret sensory information to convert it into perceptions). The colearner then constructs responses to my questions drawing upon (3) the resources of cognitive structure. These will include both the individual components of the conceptual toolkit (§1.7.2) – which act as intermediate concepts used in developing explanations and models – and the various conceptions (propositions relating concepts). This structure of conceptions will be organised to some extent, perhaps much of it as discrete minitheories, but perhaps including coherent frameworks. These may be more or less integrated (and may include partial and multiple frameworks: see figures 2.1 and 2.2). This box on the diagram represents 'stored information' that has some sort of permanence, whereas the previous box concerned the transient thought processes through which explanations may be constructed. (Osborne and Wittrock (1983, p.493) would include networks of propositions, images, episodes and intellectual skills as components of long-term memory available to a learner.)
At a simple level: a question is asked (1), which is heard and made sense of by the colearner (2), who thinks about it, calling upon ideas he or she has learnt (3), and constructs a response (4) which is expressed to the researcher (5).
However, there are other important aspects to the system. For one thing the resources of cognitive structure are not fixed, but develop. In part they will be developed by the process of answering the researcher's questions in interviews and similar situations (so that 3 is a two way process). If the colearner has developed the metacognitive awareness to become a reflective learner then he or she will actively think about the contents of cognitive structure, and deliberately develop the resources available by searching for subsuming patterns, and looking to integrate disparate parts (c.f. §2.3.10).
There are also subconscious processes, that are not well understood but are assumed to supplement the conscious thinking that the colearner undertakes. So that some aspects of the researcher's questions (for example) may not reach conscious awareness, but may still influence conscious thinking (7) through other levels of processing that access some of the sensory information (6) filtered from conscious awareness. These subconscious processes can also feedback into the colearner's behaviour – phrasing and tone of responses perhaps – without deliberate conscious control (6).
At least as important are the various subconscious processes which monitor cognitive structure and process its content to develop it into a more useful set of models of the world (3). This poorly understood phenomena explains how learning goes on over months even when there is no conscious recall of material, and explains the many cases of scientific discoveries made when the scientists were not thinking about the topic, and the folk-wisdom of 'sleeping on' a problem. This particular phenomena will be discussed in chapter 4 in the context of the processes by which the researcher makes sense of qualitative data collected from learners to construct models of their understanding. In particular the means by which analytical categories are induced from data will be considered (§4.2).
A final aspect of the model is that although it is assumed that those aspects of cognitive structure which are understood as complexes of conceptions and kits of conceptual tools are open to introspection, the construction of these resources through conscious and subconscious 'thinking', and the construction of explanations etc. from them, are mediated by Gestalts which channel the thinking process without the learner being aware.
Although this model is meant to describe the system of the researcher and colearner, it must be remembered that throughout the research the colearner is interacting with peers, text books, other teachers, news media, folk-knowledge, etc. (8), and entering into to various other worlds of discourse apart from with the researcher. These interactions are beyond the researcher's control, and mostly occur without being observed by the researcher.
In terms of this model, the aim of the research process is to construct a representation of some of the contents of the box labelled 'resources of cognitive structure' within the researcher's own cognitive structure. As can be seen from the model this process is mediated, and thus the information distorted, by the various steps in the model (and the corresponding steps to 2, 3, 4, 6 and 7 within the researcher himself). The process is also complicated by the changing nature of the colearner's cognitive structure itself, as a result of
- the processes by which his conceptual system would be developed in the absence of new information;
- the influence of his course and interactions with various aspects of his environment; and
- the particular influence of the researcher's questions and tasks.
The development of a colearner's understanding of the focal topic (chemical bonding) will involve
- adding more tools to the conceptual toolkit;
- developing more sophisticated tools which subsume existing tools;
- learning to apply conceptual tools in a wider range of (valid) problem contexts;
- arranging conceptions into more coherent complexes;
- integrating disparate complexes of conceptions into coherent overarching schemes;
and will be demonstrated by a greater ability to answer questions in terms of the accepted models and explanations of curriculum science.
