How to reliably measure working memory capacity?

Question

The capacity of human working memory is subject to individual differences. The number of items or chunks a person can retain in their memory may, have profound effects on a variety of other cognitive processes, such as problem solving.

I have come across a couple of ways to measure working memory such as the delayed match-to-sample task, Dual N-back task or the Wechsler Memory Scale. For the latter I have been unable to find out whether it offers an estimate of the number of items which can be held in working memory. This, however, is precisely what I am interested in. I know that the capacity measured using verbal stimuli is affected by a variety of features of the material.

Therefore, I'm wondering what are the most commonly used methods to measure an healthy individual's working memory capacity. Also, is there a reliable measure that can be easily used in online research? I am also interested in measures suitable for laboratory research, though. Finally, are abstract visual material and digits more "pure" measures than verbal ones?

I am especially interested in standardized procedures with known reliability measuring working memory and giving an estimate of the number of chunks an individual can retain.

Answer 1

First of all, the existence of a working memory is a theory, not a proven fact. The working memory is a concept of the information processing theory that likens the human brain to a computer. Other theories don't necessarily have an equivalent for this concept.

Second, there are different theories regarding how the working memory is made up: one single store (e.g. Atkinson & Shiffrin) or several systems (e.g. Baddeley & Hitch). Depending on what you believe the working memory is, you might have to measure it (or its parts) in different ways.

Traditionally the capacity of the working memory is measured by asking subjects to remember a list of items and then counting how many they can repeat. This is a standard subtest in many intelligence tests, e.g. the WAIS. The average (!) capacity for remembering letters or numbers has been shown to be 7 +/- 2 . Different experiments find different numbers, but they all range between 3 and 9 (as far as I know). Regular tests stop at nine items, and it is very rare that someone can regularly hold more items in his short term memory after hearing them once only (telephone numbers are stored in long term memory after being learned).

Measuring the capacity for non-linguistic visual, non-verbal auditory, olfactory, tactile etc. stimuli is extremely difficult, because as yet scientists don't even begin to understand how these stimuli are percieved, processed and represented in the brain. The only comprehensible and meaningful measurements are made of the memory of linguistic items like letters or words or visual representations of objects (like a page with a number of drawn objects).

Answer 2

What's answer above summarises many of the first things I would have said.

In addition, you may also want to note that working memory references the idea of a computer, which implies that storage and processing are separable. In fact, storage and processing occur using the same mechanisms in the brain, meaning that a simple concept of working memory is hard to credit.

Also, there is a distinction between the number of objects one can simply remember for a space of time and the number of objects one can operate on at the same time in parallel. This number changes across modalities, so that for audition, you can typically process a single audio stream at a time, however for vision, you can clearly track multiple objects across space (and in fact, if you know anything about the blind spot or peripheral acuity, the brain does a lot of that tracking on inference rather than actual perceived objects). For a good reference on processing vs simple storage and how those concepts dissociate, see Measuring the Influence of Complexity on Relational Reasoning by Birney, Damian P; Halford, Graeme S; Andrews, Glenda.

Another important issue is that WM usually refers to a specific sort of processing that occurs consciously. But behind those straightforward stores that we have conscious access to lie a vast complexity of sub conscious processing in each modality. The way that the brain tracks objects in the visual system for example, is relatively well understood, but far more complex than these higher level cognitive concepts of working memory that people talk about so nonchalantly. I'm taking your definition of working memory to be the conscious stuff, but there is a great deal of what I would term sub conscious working memory behind these processes too, that I'm not mentioning here for brevity.

I guess the key message is that there are as many measures of WM as there are definitions of it, and there are a substantial number of studies already examining the effect of different stimuli on ability to recall items. As well as the ability to recall however, WM also implies ability to operate on the items recalled (we might think of this in terms of bandwidth vs storage), which is something that perhaps you may have missed in simply talking about capacity.

Answer 3

Wilhelm 2013 What is working memory capacity, and how can we measure it?

Complex span tasks (Cspan)

reading span task (Kane et al. 2004)

operation span task

The rotation span task (Shah & Miyake 1996, adapted from Kane et al. 2004)

Updating tasks (Updating) (e.g., Miyake et al., 2000)

The verbal updating task

In the numerical updating task

the spatial-figural updating task

Recall 1-back (RNb) (Dobbs and Rule, 1989)

Binding tasks (Binding)

the letter-color binding task

the word-number binding task

Secondary memory tasks (SM)

the word-word SM task

the letter-position SM task

Reasoning tasks (Gf)

the fluid intelligence section of the Berlin Test of Fluid and Crystallized

Intelligence (BEFKI; Wilhelm et al., 2013).

Tasks measuring response inhibition (Inhibition)

the Eriksen Flanker (E) task

the Simon (S) task.