Rethinking Memory

New research offers a different model for how we remember. 

By Matthew Kelley

What do you know about memory? Over the years, I have asked this question to countless students, friends, and colleagues. Regardless of their familiarity with psychology, I am always struck by the consistency of their responses. Typically, they distinguish between short and long-term memory, that is, between temporary storage that keeps information in your mind for a few minutes versus a more permanent storehouse. They suggest that information is remembered best when rehearsed or connected to important information that already exists in memory. These ideas strongly resemble what is known by psychologists as the “standard” model of memory.

This model suggests that, upon entering short-term memory, a memory trace will immediately begin to fade away, much like a tablet of Alka-Seltzer in water. If too much time passes, the information will be irretrievably lost. Information in long-term memory, however, is never truly lost or forgotten. Memories here are assumed to be permanent and accurate representations of past events that lie dormant until we recall them. Long-term memory failure occurs when information is inaccessible, which often results from competition among similar memories or from inadequate search strategies.   

The popularity and widespread acceptance of this model stems in part because these descriptions ring true. Who hasn’t felt the telephone number you need to remember fade right out of your memory or that the information exists in your memory but you just can’t access it at the moment? 

Despite its intuitive appeal, a growing body of research has begun to chip away at the standard model. Research has suggested that people do not forget because their memories decay, that rehearsal is an ineffective strategy for maintaining memories over long periods, and no fundamental difference exists between short- and long-term memory. More theorists are arguing for a “unitary” model, in which all memories are governed by the same general principles, regardless of their time course. Indeed, much of my own research is aimed at discovering the general principles that govern memory over both short and long time intervals. 

According to this view, memories are not “stored,” as a computer might accurately store information. Instead, memories are active reconstructions of past events that may be colored by our current thoughts, emotions, attitudes, and experiences. In other words, when we recall that sweet memory of our seventh birthday party, we do not recall an exact replica of that event. We reconstruct what probably happened. Our memory will contain authentic information from the actual event, as well as general information about what usually happens at birthday parties and thoughts or perceptions that occurred well after the event took place.  

The unitary model also suggests that all memories are driven by constellations of retrieval cues, which are the details we use to recall a unique memory. Some researchers draw an analogy between memory retrieval and the mathematical concept of linear regression, in which an equation predicts an unknown value by using a variety of known values. When the equation and predictors are sound, the predicted value will closely resemble the actual value. 

For example, imagine trying to remember that seventh birthday party. We generate retrieval cues, such as a summer party in 1981 with chocolate cake, and plug them into a “memory equation.” The equation provides us with a reconstructed—or predicted—memory that should resemble the actual event we experienced, if the cues were sound.

Given the shortcomings of the standard model, and the ability for the unitary model to account for a broad array of memory phenomenon, I wonder why we cling to a clearly deficient model of memory. Is the standard model easier to comprehend or are we comforted by the idea that our memories are accurate representations of past events?

Whatever the reasons, there are costs associated with subscribing to an outdated model. Our own knowledge about how memory works exerts substantial influence over our reasoning and decision-making in everyday life. Case in point, consider how differently you might interpret the courtroom testimony of an eyewitness to a crime. From a “standard” view, one might unconditionally accept the eyewitness account as accurate and unchanging, whereas from a “unitary” view, one might regard the account with a more healthy skepticism.

It is time that we start rethinking and reevaluating what we know about memory so that we can develop a richer understanding of this fundamental process.

Matthew Kelley is an assistant professor of psychology at Lake Forest College.