This post is by Sarah Robins (pictured above), an Assistant Professor of Philosophy and affiliate member of the Cognitive and Brain Sciences Program in Psychology at the University of Kansas. Her research is at the intersection of philosophy and psychology, with a primary focus on memory. In this post she summarises her recent paper ‘Misremembering’, published in Philosophical Psychology.
Thanks, Ema, for the invitation to talk about my recent paper ‘Misremembering’ with Imperfect Cognitions readers.
The paper began from my fascination with one of the most common experimental techniques for eliciting memory errors: the Deese-Roediger-McDermott, or DRM, paradigm (Deese 1959; Roediger and McDermott 1995). I am fascinated because these errors display a blend of success and failure (on which I will elaborate on below). In the paper, I argue that they are best viewed as a distinct type of error, misremembering. I go on to argue that we lack a theory of memory that can explain misremembering. I divide theories of memory into two broad groups: traditional Archival accounts and contemporary Constructive ones. Each is insensitive to the explanatory demands of misremembering errors, but in a distinct way. In short, Archival accounts do well at explaining memory’s successes and Constructive accounts do well at explaining memory’s failures. But since misremembering errors involve success and failure, they present a challenge to both.
To see the point, let us turn first to the DRM. In this paradigm, participants are presented with a list of related items to memorize and then are later quizzed about which items were on the list. Here’s an example (you are welcome to try this out on a friend! The standard results are fairly easy to obtain):
After a short break, participants are asked three types of recognition questions:
- Was JUICE on the list?
- Was RADIO on the list?
- Was FRUIT on the list?
Most participants do well with questions 1 and 2; JUICE was clearly on the list and RADIO was clearly not. The difficulty comes with question 3, about words that are related to those on the list but were not actually included. Many participants claim to recognize FRUIT as being on the list. They report recognizing it as often as words that were on the list, like JUICE, and in some cases, will even provide details of what went through their minds when they heard this word (which, of course, is a false memory because the word was never presented). Results from the DRM are robust; they are well replicated across changes in stimuli, retention interval, and even warnings against making such errors. The DRM results are so well-established, in fact, that this paradigm is often used as a baseline against which to measure the efficacy of other experimental manipulations.
Hopefully it is now clear why I characterize these errors as a blend of success and failure. Participants in the DRM appear to both forget and remember what was on the list. When a participant claims to recognize FRUIT, she errs. But to claim that FRUIT and JUICE were on the list, while RADIO was not, relies on successful retention of the list-learning event.
I argue that DRM errors are instances of misremembering, which I define as follows:
Misremembering is a memory error that relies on successful retention of the targeted event. When a person misremembers, her report is inaccurate and yet the error is explicable only on the assumption that she has retained information from the event that her representation mischaracterizes.
Misremembering errors do not require a laboratory. They are familiar from everyday experience. When reflecting on a past party, you might recall that your aunt arrived late when it was actually your cousin, or that this was the occasion when your aunt told a particularly funny story, when this actually occurred at a different gathering.
With a clear understanding of these errors to hand, my next task in the paper is to see whether either of the two standard approaches to memory—the Archival View and the Constructive View—can explain these errors.
The Archival View is the traditional, default view of memory, according to which memory is a preservative capacity directed at the past. The view is shaped around the assumption that—at least some of the time—we form and retain discrete representations of our past experiences. This account does well at capturing what happens in cases of remembering, but Archivalists struggle to accommodate DRM errors because they are at a loss to explain how one could retain information from a past event and yet fail to produce an accurate representation of that event when remembering. If I have stored the list of words, so I can recognize JUICE and reject RADIO, then why would I also then err and “recognize” FRUIT?
Proponents of the Constructive View argue that memory is a capacity for building (i.e., constructing) representations of past events. Memory still preserves information from the past, but this information is retained in a generalized way; there is no retention of information from particular past events. This enables the Constructivist to explain why memory errors so often involve errors and distortions. But appeal to the various influences on retrieval can only explain so much. The Constructive approach leaves out a key influence on misremembering errors like those of the DRM: information retained from the particular past event. The relationship between FRUIT and JUICE influences participant errors on the task, but so does the particular past event of learning the list.
In the paper, I provide more detail about versions of Constructivism and explore many alternative ways that proponents of each view could explain the DRM effect in particular and misremembering in general. I conclude that neither the Archival or Constructive account can provide a plausible account of these errors.
Fortunately, the explanatory limitations of the Archival and Constructive Views are complementary, which suggests a way forward. I think that explaining misremembering will require a hybrid theory of memory, combining the Archival View’s commitment to retention of discrete memory traces with the Constructive View’s approach to retrieval. In the paper, I only sketch the beginnings of what such an account might look like. Providing a detailed account of this hybrid view is the task of my current and future work.