Welcome to Memory Mondays, where I read a textbook on memory and talk about what I Iearned. If you like your cognitive psychology neatly summarized, with a healthy dose of unnecessary commentary and bad humour, this is the series for you!
Look at your desk.
Not at your desk? Okay, use your visual long-term memory to picture your desk in your mind.
If you’ve got a lot going on, it is probably pretty cluttered. Notebooks to check, tests to grade, Post-Its with reminders, your (dirty) coffee mug, a couple teaching books, and, somewhere in there, your computer screen and keyboard. The more you’ve got going on, the bigger the workspace you need to stop it from getting out of hand.
Working memory is essentially your desk, but for your mind.
Admittedly, the analogy quickly falls apart if you work at a paperless school, or are organized enough to quickly hand back student work and regularly clean your coffee mug, but the important idea is this: your working memory is your mental workspace.
It’s where you store the things you are thinking about.
It’s where you do your thinking.
Effective use of this space is important for a wide range of cognitive activities, and thus is instrumental in the performance of our students .
So let’s get started.
Atkinson and Shiffrin’s Modal Model
Before we look at the multicomponent model, elements of which were discussed last week, let’s look at the main model that preceded it: Atkinson and Shiffrin’s modal model, developed in the late 1960s.
In this model, information is first processed by sensory memory, then moves into short-term storage, which serves as working memory and also feeds information in and out of long-term memory.
A few problems became apparent with this model. First, the model proposed that merely holding information in the short-term store for long enough would ensure transfer to long-term memory and thus learning. Craik and Lockhart challenged this view in the early 1970s with their levels of processing theory, which asserted that learning is a product of how the information is processed, rather than time spent. As well, the modal model suggests that a deficit in short-term store would create a deficit in working memory, as they are the same. Studies of individuals with greatly impaired short-term store, as well as a study where a concurrent digit span task was used to impair short-term store while a reasoning task was completed, show that this is not the case. Working memory is not wholly reliant on short-term store.
With these shortcomings of the modal model, a new model was needed.
Baddeley and Hitch’s Multicomponent Model
Baddeley and Hitch proposed an alternative to the modal model in 1974. It divides working memory into four parts: phonological loop, visuo-spatial sketchpad, central executive, and episodic buffer.
The first two parts were addressed in the previous chapter, as they mostly serve as storage, but there are some interesting aspects of both related to the ‘working’ part of working memory. The phonological loop can help moderate task switching through subvocalized self-instruction: essentially, talking to yourself inside your mind about what is happening while you are trying to multi-task. The verbal part of memory can help with visuo-spatial, and the visual with verbal. For example, using visual imagery can help people memorize a list of word pairs, and people skilled with an abacus can imagine the device to help increase their digit span. These two stores are separate, but still connected.
Of most interest to me was the research on why we have a phonological loop. The visuo-spatial sketchpad supports us in the planning and carrying out of action, but the purpose of the phonological loop is less clear. People with severely impaired digit spans still appear to have normal lives. So what is the purpose of the phonological loop?
The answer appears to lie in language acquisition. Studies on adults learning a foreign language and young children learning their native language show a relationship between verbal short-term memory and vocabulary acquisition. For me, this partially answers a question that has been sitting at the back of my mind for a while. As an international educator, I teach a student population that is close to 100% English language learners and has a wide range of English proficiency, from far below grade level to native-like. I have considered different explanations for this variability, such as time in an English-language school, motivation, and quality of instruction at their previous schools, but differences in verbal short-term memory add another piece to the puzzle. Basically, reading this chapter was a small eureka moment for me.
The central executive is more about attention, rather than about memory storage. It has the capacity to direct attention to a task, or to divide attention between two or more tasks. It is important to note that the latter only works well when the tasks are simple. That is why talking on a phone becomes a problem, not during routine driving situations, but more complex situations where a driver needs their judgement and full attention to avoid an accident.
The episodic buffer was added to the model much later, to account for how working memory interacts with long-term memory. It is a storage system, but it is multidimensional, holding verbal, visual and spatial information, and, most importantly, binding these different dimensions together; it pulls information from short-term stores, long-term memory, and perception (sensory memory).
Alternative Theories for Working Memory
The text discusses three alternative theories to Baddeley and Hitch’s model (or, at least, I have three in my notes from the chapter), but I’m only going to focus on the one that I feel offers the best challenge to the multicomponent model: Cowan’s embedded processes model. In other words, it is the model I understand the best of the three.
All of the alternative models are described as top-down, and seem to me to focus on what would be the functions of the central executive in the multicomponent model: attention in Cowan’s model, blocking out irrelevant information in the inhibitory model, and dividing attention between tasks in the time-based resource-sharing model. In contrast, the multicomponent model is considered bottom-up, as it started with the study of verbal short-term memory.
For Cowan, working memory is about activation of long-term memory, and it is activated through attention. This activation decays unless maintained through rehearsal or continued attention. Cowan argues that working memory has a capacity for four ‘chunks,’ which I guess is what we can maintain attention on.
Usually I find diagrams in cognitive psychology of little help in clarifying a concept, but the diagrams for Cowan’s model, such as this one or this one, did improve my understanding, so take a look and see if they will help you as well.
I would have liked Cowan’s model to be described in more detail, as my understanding of it is still pretty shaky; Baddeley described it as “influential,” and it is the only other model of working memory mentioned in the cognitive psychology textbook I read previously, so it seems to be the main alternative to Baddeley and Hitch’s model.
Because the implications for education are my driving motivation, this section was a bit of a disappointment. Seeing as working memory is our mental workspace, there are lots of questions to consider. How can we design lessons to avoid overloading working memory? How does working memory affect critical thinking skills? Can we improve working memory? This section seemed more concerned with tests that can identify students with low working memory capacity.
In defense of the text, it does partially answer the latter two questions. When discussing individual differences in working memory, correlations with language comprehension, prose composition, obeying complex instruction, performance in programming courses, and IQ, are all highlighted and various research studies supporting these findings are cited (a handy reading list for me later).
Baddeley also discusses the question of whether working memory can be improved through training. It was a much more optimistic discussion than I expected. Based on what I have read previously on edu-blogs, I assumed the answer was more-or-less “no.” The meta-analyses of studies discussed by Baddeley were varied in their interpretation of research. One was more pessimistic, pointing out how gains from working memory training programs tended to be short-term. While the training programs do show generalization to other working memory tasks, there is little evidence of generalization to non-laboratory tasks; the more optimistic view is that generalization to the real world needs further, specific, training, as is often needed in psychological and medical rehabilitation as well.
While I felt that the discussion of working memory in education was brief, I also think that the discussion needs to be in the context of long-term memory, and may materialize in later chapters.
Next week: learning!