Remembering and forgetting

Memory is a key concept related learning – but some common-sense ideas about memory are found to be questionable at best.

This short extract is taken from a research paper: Taber, K. S. (2003) Lost without trace or not brought to mind? – a case study of remembering and forgetting of college science, Chemistry Education: Research and Practice, 4 (3), pp.249-277. [Free access]

Remembering and forgetting

"Despite there being a considerable body of work exploring aspects of learning in science, there seems to be little – if any – research primarily focusing on the 'forgetting' of science learning. However, the work of Gauld (1988, 1989) is of particular interest here. Gauld reports a study of classroom learning about electric circuits, where students' ideas were elicited before the classroom presentation, soon after learning, and then again after a delay of some months. Gauld's work derived from a study by Cosgrove and Osborne (1995), who had devised a teaching sequence designed to bring about conceptual change by providing critical experiments to challenge students existing mental models of circuits. Comparisons between students' ideas before and soon after the sequence of lessons showed significant shifts in thinking towards the more scientific model: but a year later many of the students seemed to have reverted to their prior thinking.

Gauld found a similar effect in the follow-up study, which this time included interviews with students 3-4 months after the sequence of lessons. Part of the teaching sequence involved students predicting the relative sizes of ammeter readings in circuits. So, for example, a mental model of current being 'used-up' around a circuit would lead to a prediction of decreasing readings. The demonstration that all the ammeters would show the same reading would provide a challenge to this model, and encourage students to adopt the scientific model where current is constant around the circuit.

What is of particular interest to the theme of the present paper [ Lost without trace or not brought to mind? – a case study of remembering and forgetting of college science] is that – despite the efficacy of the evidence in changing students' minds in the short term – Gauld found that in the delayed interviews some students justified a model of current being used up in terms of their recollection that ammeter readings decreased around the circuit. In other words, although the primacy of the evidence seemed to bring about conceptual change immediately after teaching, in time the memory of the evidence became altered to fit the preferred mental model.

So here we see that not only are perceptions of classroom demonstrations and experiments often distorted by students' existing conceptual frameworks (Driver, 1983), something known in memory research as 'encoding distortion', but even when initial interpretation of experiments are accurate, they may later give rise to false memories. This finding is not surprising in the light of memory research where eyewitness testimony has been found to be notoriously unreliable.

A different feature of memory has been studied in work deriving from the information-processing tradition, in relation to problem-solving in science. So … work by Tsaparlis (1998) has explored the way that problem-difficulty in organic chemistry depends upon the number of different chunks of information that need to be activated by a learner in order to successfully solve problems. His results suggest that information that is cued and so recalled in the context of some problems may not be accessed by the same learner in other problem contexts that include the same cues but require more information to be recalled or considered. This was found even though the problems used were considered to be well within the nominal capacity of the student's working memory. Such research is very significant for the way students are assessed. The items used by Tsaparlis were authentic chemistry question of a type commonly incorporated in formal tests and examinations – responses to which are commonly taken as prima facie evidence of whether learning has taken place. However, the relationship between (i) student learning, (ii) examination questions, and (iii) students' responses, is in reality much more complex than this (Taber, in press).

Various aspects of memory have received attention in psychology, although it has been suggested that – despite the topic being the subject of much literature – our current understanding is in many aspects rather provisional (Ramachandran & Blakeslee, 1998). Here I provide a, necessarily very brief, review of current thinking about those aspects of memory relevant to the present study (Guildford, 1972; Sacks, 1985; Luria, 1987; Parkin, 1987; Ellis & Hunt, 1989; Baddeley, 1990; Eysenck & Keane, 1990; Rose, 1992; Anderson, 1995a, b; Erdelyi, 1996; Greenfield, 1997; Kellogg, 1997; Carter, 1998; Catania, 1998; Dawson, 1998; Pear, 2001; Sousa, 2001).

It is not known for certain whether once something is committed to long-term memory the trace can sometimes ëfadeí away, but it is widely believed among memory workers that complete extinction of such traces is unlikely. Failures to remember information that has been learnt are generally considered to be due to an inability to access the memories (i.e., to activate the trace), so that the memory is considered available but inaccessible rather than unavailable. Memories not brought to mind for many years have been known to be triggered by drugs, medical conditions such as migraine or epileptic events, or direct electrical stimulation of the cortex during surgery, sometimes much to the surprise of the individual experiencing the reminiscence.

Accessing memories (in the absence of chemical, surgical or traumatic intervention) requires suitable cues to activate the memory trace. Although some memories seem to be much more readily accessed than others, much recall of information is highly dependent upon the cues available. Such cues can include appropriate category labels, or other information stored at the same time (whether conceptually related or incidental). In view of this, forgetting should not be seen as absolute, but rather seems to be best considered as not recalling a memory in a particular context – at a particular time, with the subject in a particular mood or environment (perhaps matching that when learning took place), in response to a particular question etc.

Just as forgetting is not an 'all or nothing' phenomenon, so the formation of a memory should not be considered as an event, but rather as a process. Current thinking is that memories continue to be developed over a long time span. The process of 'consolidation' takes place over months and years (largely during sleep) and results in memories being further integrated with other information in cognitive structure. Indeed, there is a phenomenon known as hypermnesia whereby greater recall of information becomes possible after increasing periods since initial 'learning'.

As well as this subconscious process of forming connections (and therefore changing) memories, it is also believed that each time a memory is activated and called to mind (i.e., transferred from long-term store into working memory) it is both reinforced and altered. So, memories can be strengthened by the conscious process of reviewing prior learning and relating this to new material, but even without this deliberately directed activity similar evolution of memories is taking place outside of conscious control.

One consequence of the way that consolidation is a long-term process is the phenomenon of interference. When new learning is similar enough to poorly consolidated prior learning it is possible for the two sets of information to become conflated, so that cues that might have activated the earlier memory may instead activate the newer information. Conversely, newer learning that is not reinforced may become indistinguishable from previously consolidated learning. This means that checking that students can clearly distinguish novel concepts from similar notions learnt previously soon after the presentation of new material is not sufficient to ensure that the ideas will remain distinct over a longer time span.

Just as we are not aware how our memories may have been altered since initial learning, we also usually fail to appreciate the degree to which memory is a reconstructive process. When the available cues are sufficient to activate only some of the relevant information we tend to fill in the details ('confabulate') to give the memory an overall coherence, so that some of what we 'remember'  is actually inference based on making sense of the information successfully accessed. As each act of recall is believed to act as rehearsal to strengthen the memory trace such confabulations can become consolidated into memory (so that on future occasions the 'false' memories are part of the information genuinely recalled).

In the light of such a view of memory it seems less surprising that some of Gauld's interviewees confidently reported false records of classroom demonstrations that were meant to act as critical experiments. Indeed, such a perspective on memory helps explain why it is so common for students who attended, and attended to, their science classes to produce unsatisfactory answers to assessment questions on topics they have studied. Even when course material is initially understood as intended – which is by no means always the case – this does not ensure it will be activated and recalled in an unadulterated form when cued by the assessment question.

Where students from a cognitive psychology class were tested for their memory of the class over an extended period it was estimated that they could recall about a third of the material after about ten years. However the 'forgetting' was not uniform: rather there was a quick decline over a period of about forty months, after which memory seemed to stabilise, so that material still accessible at this time was not later 'forgotten' (Kellogg, 1995). …"