The problem with gist

The term ‘gist memory’ usually refers to less detailed long-term memories, often (but not exclusively) episodic memories. However, our semantic memories often rely on these memories of events as we use situational cues to help us recall them. It then follows that errors here can have a knock on effect.

When we read a book, for example, we don’t appear to retain all the information from it, but we do get a general idea of its content. If you were to ask me about Cormac McCarthy’s The Road (I read it quite recently), I’d tell you it’s about a father and son journeying through a post-apocalyptic America and that they encounter some very unsavoury characters along the way. I could tell you how it ends (don’t worry, I won’t) but I won’t be able to go into detail about every incident in the story. The same is true for the film based on the book. What I have is the gist of the story, but not a blow-by-blow account. This notion of gist memory has been known for some time. George Stout was an early British psychologist and author of the highly influential Manual of Psychology (first published in 1898). In his manual he describes how a student, studying Euclid for the first time, will initially attempt to learn it by heart. However, when the student had completed the work, all that will remain is the ‘general method of proof’ (Stout, 1901 p.454), much of the detail will have been forgotten. 

The Problem with Long-term Memory

What we can’t know for sure is when detail becomes gist. The general view is that the detail will gradually fade, leaving behind only the residue of what was experienced, the details we believe to be most relevant or the most interesting. However, the fallibility of long-term memory can also result in important detail being stripped from our recollection or, more damaging perhaps, a memory that bears little resemblance to the one experienced.

Say I was to give you a list of medical related words, for example:

nurse

patient

hospital

ambulance

If I included a delay or an interference task between the learning element of the task (the encoding stage) and the recall stage, and then asked you to identify the words from the following list:

computer

nurse

bicycle

newspaper

hospital

doctor

school

ambulance

you would be statistically more likely to include the word doctor along with nurse, patient, hospital and ambulance. Crucially, your confidence that doctor was part of the original list would be very high, even though it wasn’t included in the initial list. This is a type of test (known as the DRM task after its originators; Deese, Roediger and McDermott) often used to induce false memories, that is, recollections of which we are very confident despite them having never taken place. Ebbinghaus was certainly correct in his assumption that the passing of time plays a vital role in what we remember and what we forget, but time also plays a part in how we remember. Cognitive psychologists working within the criminal justice system have long been aware that long-term memory is far too error prone for eye-witness testimony to be even close to accurate. For example, for decades Elizabeth Loftus and her team at the University of Washington have been gathering evidence from hundreds of studies that find people simply don’t recall events in the way we think they do. Loftus has also demonstrated again and again how simple it is to take advantage of memory fallibility by creating false memories in many of her volunteers.

It’s all a bit fuzzy

Many of our recollections, therefore, appear more as vague resemblances than accurate memories. However, Valerie Reyna and Charles Brainerd think it’s a little more complicated than that. Reyna and Brainerd suggest memory traces are the product of two different processes; gist memory and verbatim memory, and that both exist simultaneously for the same event. Fuzzy Trace Theory (sometimes referred as Dual Trace Theory), was developed to explain how false memories might arise and how this impacts the criminal justice system in terms of eye-witness testimony. But it can also provide some insights into how what we believe we’ve learned can turn out to be, not only inaccurate, but also really way off the mark.

I’ll illustrate these two types of memory with a very simple example. As I’m sitting here typing I’ve just remembered that I’ve left a drink and a snack in the kitchen (when I’m working I’m generally fuelled by tea, coffee and a plentiful supply of biscuits). This description, that I’ve left my drink and snack in the kitchen, is gist memory, devoid of any specific detail. Nevertheless, there are more details I could divulge; the drink is tea (and Yorkshire Tea at that) and the snack is biscuits (Jammie Dodgers, because in my head I’m still seven). The first description involves non-recollection retrieval and focuses on the semantic details rather than the surface details, while the second is recollective memory, where I’ve mentally reinstated the contextual features of my memory with all the details attached; the Yorkshire tea in my favourite Doctor Who mug that yells Exterminate when I pick it up and there is a plate with four Jammies Dodgers (well, actually, three and half because I took a bite out of one while I was waiting for the kettle to boil). Both the mug of tea and the plate with biscuits are located just in front of the kettle. According to Fuzzy Trace Theory, both these gist and verbatim memories form at the same time, but we can store the gist memory after only processing a fraction of its details (Draine & Greenwald, 1998). We then have access to two records of the same experience, the detailed verbatim one (the actual event), and the more superficial gist one (the meaning of the event). 

Recounting my memory will always contain a mixture of verbatim and gist retrieval, so sometimes I’ll say, I’ve left my tea and biscuits in the kitchen, that is, I’ve reconstructed the event from gist. At other times (although much less likely), I’ll describe the event in much more detail, down to the type of biscuits and the Doctor Who mug. The problem is that, because gist becomes mixed up with verbatim, I may well misremember, creating (in effect) a false memory. My gist memory will certainly contain some true memories, but it’ll also involves the false ones. The gist memory merely reminds me that I’ve left my drink and a snack in the kitchen, so I might recall coffee instead of tea or Oreos instead of Jammie Dodgers. I’m assuming the tea is in my Doctor Who mug, but it might not be. All or some of these recollections might be wrong, or they all might be accurate. If you were to escort me into a darkened room, shine a light directly into my face and interrogate me about what I left in the kitchen, the questions you ask, and how you ask them, are going to influence what I genuinely remember, what I’ve forgotten and what I truly believe to be correct, but actually isn’t.

Your questions might act as cues that nudge my memory, but they might also lead me astray:

Did you make your coffee in the Doctor Who mug?

Well, I did make a drink in it, but now I’m not sure if it was tea or coffee because you seem pretty confident it was the latter.

However, my verbatim memory may well fight back:

No, I know it was tea because I remember putting the bags in the pot (yes, I’m old enough to still have a teapot).

My responses will also depend on which of the traces are stronger and how long has passed since the event and the interrogation, but over time, the verbatim memory will fade more rapidly than the gist memory (Kintsch and Mangalath, 2011). If I had recalled making a cup of coffee rather than tea, this would represent a false memory.

Individual Differences

How verbatim and gist memory mix depends, in part, on our age. Both systems do improve between childhood and adulthood; in early development there’s an asymmetrical relationship, in that gist memory evolves more gradually than verbatim memory. However, in late adulthood, verbatim memory declines while gist memory is largely spared. It also appears that some people are more susceptible to the creation of false memories through verbatim-gist memory interaction. This is particularly the case for people with clinical conditions such as PTSD, but it would also appear that adults displaying lower levels of intelligence are also susceptible, as are elderly adults. Interestingly, although not particularly surprising, adults who hold beliefs that violate established scientific facts (such as climate change denialists and other conspiracy theorists) are also more likely to fall victim to false memories.

Contrary to early descriptions of false memories that considered them as somewhat vague and and less vivid than real memories (e.g. Conway, Collins, Gathercole and Anderson, 1996), more recent accounts have found them to be virtually indistinguishable (see, for example, Lampinen, Neuschatz and Payne, 1998). According to the Phantom Recollection Principle, false memories are often vivid. According to Arndt (2012), when people remember by re-constructively processing gist, they can recover realistic contextualised details that make false memories seem real.

Fuzzy Trace Theory might explain why you only have a vague idea or what you learned at school, because verbatim memory is going to fade much quicker than gist memory, so all you’re left with is a vague recollection of what you, for while, seemed to know in detail. There are, of course, other explanations, but some of them are much less plausible, such as the Zeigarnik effect. Perhaps just as worrying (at least in terms of learning) is that we retain the gist of something we’ve learned but the details are inaccurate, potentially leading to errors and misconceptions.