How fat is your memory?

A chemical analogy for working memory

Keith S. Taber

This posting has nothing to do with the chemical composition of your brain (where lipids do play an important part), nor about diet – such as those claims of the merits of 'omega-3 fatty acids' in a healthy diet.

Rather, I am going to suggest how a chemical structure can provide an analogy for thinking about working memory.

In a sense, working memory is a bit like triglyceride structure (which is only a useful comparison for those who already know about the chemistry being referenced)

Analogies are used in teaching and in science communication to help 'make the unfamiliar familiar', to show someone that something they do not (yet) know about is actually, in some sense at least, a bit like something they are already familiar with. In an analogy, there is a mapping between some aspect(s) of the structure of the target ideas and the structure of the familiar phenomenon or idea being offered as an analogue. Such teaching analogies can be useful to the extent that someone is indeed highly familiar with the 'analogue' (and more so than with the target knowledge being communicated); that there is a helpful mapping across between the analogue and the target; and that comparison is clearly explained (making clear which features of the analogue are relevant, and how).

So, the similarly is in terms of conceptual structures. Consider the figure above, which suggests there are similarities between aspects of the concept of working memory and aspects of the concept of triglyceride structure. In this case the analogy is at quite an abstract level – so is only likely to be useful for more advanced learners (such as science graduates preparing for teaching for example).

In relation to science, we might distinguish between several classes of analogy. In teaching we are most likely to be explaining some target scientific idea in terms of an everyday idea or phenomenon. However, sometimes one scientific idea which is already well-established is used as the analogue by which to explain about a less familiar scientific idea. This can happen in science teaching or science communication to the public – but can also be employed by scientists themselves when communication new ideas to their peers. It is also possible that sometimes a scientific idea may be useful as an analogue for explaining some target idea from outside science (as long, of course, as the science is familiar to the audience for the analogy).

Read about science analogies

An analogy for science teachers

My professional life has basically encompassed teaching about three broad areas – teaching natural science (mainly chemistry and physics) to school and college learners; teaching educational ideas to those preparing for school teaching; and teaching about research to those setting out on research projects.

The analogy I am discussing here came to me when preparing the manuscript for a book aimed at teachers of chemistry (an audience of readers that I can reasonably assume have a high level of chemistry knowledge) and broadly about pedagogy. So, what came to mind was an analogy from science to put across an idea about what is known as working memory.

Working memory

Working memory is the name given to the faculty or apparatus we all have to support conscious thinking – when we plan, assess and evaluate, problem solve, and so forth. It is absolutely critical to our nature as deliberate thinkers. We probably do MUCH more thinking (if you allow that term in this context, if not, say, cognitive processing) pre-consciously, so without any awareness. This is the 'thinking' [or cognitive processing] that goes on in the background, much of which is quite low level, but also includes the kind of incubation of problems that leads to those sudden insights where a solution comes to us (i.e., to our conscious awareness) in a 'flash'. It is also the basis of those intuitions that we describe as 'gut feelings' and which are often powerful (and often turn out to be correct) even though we are not sure what we are basing them on.

Yet working memory is where we do the thinking that we are aware of, and supports the stream of consciousness that is the basis of our awareness of ourselves as thinking beings. Given its importance, a very interesting finding is that although the brain potentially has a virtually inexhaustible capacity for learning new information (through so called 'long-term memory'), the working memory itself where we process material we are trying to learn and memorise has a very limited capacity. Indeed, it is often said that typical working memory capacity in a normal adult is 7±2. And some think that may be an overestimate. So, a typical person can juggle no more than about seven items in mind at once.

There is a very important question of why such an important aspect of cognition is so limited. Is there some physical factor which has limited this, or some evolution contingency that is in effect an unlucky break in human evolution? There is also the intriguing suggestion that actually this very limited capacity may have survival value and so be considered an adaptation increasing fitness.1 Whatever the reason – we have a working memory that can be considered to only have about half a dozen slots for information. (Of course, 'slots' is here a metaphor, but a useful one.)

Chunking

Each slot will take one item of information, except that we have to be careful what we mean by one 'item', as the brain acts to treat such 'items' subjectively. That is, what counts as one item for your brain may not work as one item for mine. Consider the following example:

1s22s22p63s1

How many 'items' is that string of symbols? If we consider someone who only saw this as a series of numbers and letters and who had never come across this before they would need to remember:

  • the number 1,
  • is followed by the lower-case letter s,
  • followed by the number 2,
  • which is a superscript,
  • then the number 2,
  • then the lower-case letter s,
  • then the number 2,
  • which is a superscript,

and quite likely we have exceeded memory capacity when only half way through!

But for a chemist who already knows that this particular string could be seen as the electronic structure of a sodium atom,2 this can be treated as one unit – the whole string is already available represented as an integrated structure in long-term memory form where it can be copied as 'a chunk' into working memory to occupy a single slot. So, a chemistry teacher using this information in an argument or calculation has other 'slots' for the other relevant information whereas a student may be struggling.

Triglycerides as an analogue for working memory

It struck me that an analogy that would be familiar to many chemists and science teachers is that of triglycerides which are considered esters of glycerol (with its three alcohol groups) with fatty acids. Although this class of compounds has some commonalities, there are a great many possible different specific structures (each strictly reflecting a distinct compound). What is common is the short chain of three carbons each bonded to an ester linkage (left hand figure below). However, what those ester linkages actually link to can vary. In human milk, for example, there are a great many different triglycerides (at least of the order of hundreds) comprising a wide range of fatty acids (Winter, Hoving & Muskiet, 1993).3

The triglycerides are members of a class of compounds with common features. They can be considered to be the result of glycerol (propane-1,2,3-triol) reacting with fatty acids, where the compounds formed will depend upon the specific fatty acids. The first figure uses R as a generic symbol to show the common structure. The second figure is the simplest triglyceride type structure formed when the acid is methanoic acid. If a mixture of acids is reacted with the glycerol, the side chains need not be the same – as in the third example. Actual triglycerides found in fats and oils in organisms usually have much longer chains than in this example.

In the image above, meant to be a simple representation of the structure of triglyceride molecules, the first figure has Rs to represent any of a great many possible side chains. The shortest possible structure here just has hydrogen atoms for Rs (triformin – the second figure), but more commonly there are long aliphatic chains as suggested by the third figure – although usually the chains would be even longer. In relation to diet, a key feature of interest is whether the fats consumed are saturated, or have some degree of 'unsaturation' (i.e., the double bond shown in the middle chain of the third figure) – unsaturated fats tend to be seen as more healthy, and tend to come from plant sources.

We might consider that the molecular structures consists of the common component with three 'slots' for side chains. In principle the slots could be occupied by hydrogens (hydrogen 'atoms') or chains based on any number of carbons (carbon 'atoms').4 So the total mass of a triglyceride molecule can vary considerably, as can the number of carbon centres in a molecule.

Fixed 'slots', variable content

Working memory is sometimes said to have 'slots' as well (again, to be understood metaphorically) into which information from perception or memory can be 'slotted'. We can consciously operate on the information in working memory, for example forming associations between the material in different slots. The number of slots in a person's working memory is fixed, but as information that has been well learnt can be 'chunked' into quite extensive conceptual structures, the total amount of information that can be engaged with is highly variable.

Working memory has a very limited number of 'slots' – but where extensive conceptual frameworks are already well established from prior learning a great deal of information can be engaged with as a single chunk

If student in class is keeping in mind information that is not directly related to the task in hand then this will 'use up' slots that are not then available for problem-solving or other tasks. Indeed, one of the skills someone with expert knowledge in a field has, but not novices, is determining which information available is likely to be peripheral or incidental rather that important to the task in hand, and indeed which of the important features need to be considered initially, and which can be ignored until later.

The perceived complexity of a learning task then always has to be considered in relation to the background knowledge and experience of the individual. So, at one time a person may be watching a documentary on a subject they know nothing about, in which case the information they perceive may seen unconnected, such that working memory may be occupied by very small chunks (such as individual names of unfamiliar people that are bring discussed). If that same person sits down to revise course notes they have developed over an extended time time, and have reviewed regularly, they may be bringing to mind quite extensive conceptual structures to slot into working memory. The same working memory, with the same nominal capacity, is now engaging with a much more extensive body of information.

Work cited:
  • Winter, C. H., Hoving, E. B., & Muskiet, F. A. J. (1993). Fatty acid composition of human milk triglyceride species: Possible consequences for optimal structures of infant formula triglycerides. Journal of Chromatography B: Biomedical Sciences and Applications, 616(1), 9-24. doi:https://doi.org/10.1016/0378-4347(93)80466-H
Footnotes

1 The logic here is that, because of chunking, working memory biases cognition towards what is already familiar, which may be an advantage in a context where although change is important there is a largely stable environment so that developing and then following a stable set of survival strategies is generally advantageous.

The kind of fruit that was edible yesterday is probably edible today, and the animal that attacked the group last week is best assumed to be dangerous today as well. The peer who helped us yesterday may help us again in future if we reciprocate, and the person who tried to cheat us before is best not trusted too far today.

2 There is a strong case that the familiar designation of electronic structures in terms of discrete s, p, d and f orbitals is only strictly valid for hydrogenic (single electron) species – but the model is commonly taught and used in chemical explanations relating to multi-electron atoms.

3 Strictly there are no fatty acids 'in' the triglycerol just as strictly there are no atoms in a molecule.4 I am here using economy of language which will be clear to the expert, though we risk misleading novice students if we are not careful to be precise. The triglycerides have various chain segments corresponding to a wide range of fatty acids; a wide range of fatty acids are generated by hydrolysing the triglyceride.

4 Strictly 'atomic centres', as molecules do not contain atoms, as atoms are by definition discrete structures with only one nucleus – and the atomic centres in molecules are bound into a molecular structure. Again, chemists and teachers may refer to carbon atoms in the side chain knowing this is not precisely what they mean, but we should perhaps be careful to be clear when talking to learners.