Oh dang. It happened again. I walked into a room full of people and experienced retrieval failure. It’s a birthday party and although I’ve met most people on the host’s previous birthdays, some of the guest only look vaguely familiar. A woman greets me (she knows my name) and asks me about my recent trip. Her face is familiar, as is her voice. I know that we have chatted before and I must have told her about my travel plans (and she remembers!). Now I feel embarrassed that I don’t recall who she is. I try some vague questions (And how are you? How are things at work?) but her answers do not provide the right cues that might help me retrieve useful information. Why have I forgotten this woman? And how can I prevent this sort of situation in the future?
The Psychonomic Society’s journal Memory & Cognition has published a special issue to commemorate the Atkinson and Shiffrin memory model that has become known as the Modal Model. This model, as well as later developments such as SAM and REM, is a mathematical model that explains and predicts memory performance in controlled laboratory experiments where participants perform tasks such as study a list of unrelated words which they later try to recall. Models of this type are extremely clever, and amazingly good at predicting human memory performance using just a few basic mathematical equations. Judging by the responses of our undergraduate students, however, I suspect that most people might not immediately see how such memory models bear on everyday situations such as forgetting a name.
But they do, and they might even provide us with tools to become better at retrieval.
I will focus on the SAM model. In this model, the key to successful retrieval from memory is the quality of the memory cue. In the model, the to-be-retrieved information is called the target. Going back to the party, my memory cue might be the woman’s face and the target might be her name and her relation to the party’s host. Let’s assume that she is the host’s cousin Lisa. Lisa’s face and voice might be cues to several memories, not only of our conversation last year, but also to memories of similar looking family members, or to other things that happened while I was in her presence. Moreover, there are differences between Lisa’s face in my memory and how she looks now.
In the model, remembering happens in two steps, sampling and recovery. Memories are sampled with a certain probability. This probability is based on the strength of the association between cue and target relative to the strength of the association between the cue and all memories associated to it. So, the stronger the association between Lisa’s face and my memory of her name, the more likely it is that I will sample that memory. If I have strong associations between her face and other memories, however, the less likely it is that I will sample her name.
Mensink and Raaijmakers introduced the idea that forgetting can be explained by contextual fluctuation. They proposed that when you try to retrieve something from memory, you include the current context in the memory cue. Context is a bit of a tricky concept, but let’s say context is the situation of the party, such as the host and the room, and my physical and mental state. Context is helpful as a retrieval cue because it narrows down which memories are sampled. However, between now and the last time I spoke to Lisa, I may have been in that same room on other occasions when she was not there. I may now feel very different (frustrated) from the last time I spoke to her (happy). In general, because context changes continuously, it becomes less helpful as a retrieval cue throughout time. As a result, the party room or my mood are no longer good retrieval cues for Lisa’s name, but they are retrieval cues for many other, more recent memories. So, as I am trying to come up with Lisa’s name, I recall other stuff that gets in the way of finding Lisa. Although this insight might not be really helpful by itself, it is comforting to know that we do not always forget because memories are gone, but because other, stronger memories are added.
The next step in the SAM model, after a memory has been sampled, is recovery of the memory content. Thus, while sampling may lead to the correct memory, Lisa’s name still needs to be recovered. The probability of recovery depends on the strength of the association to the cue, but now the other, competing memories are no longer relevant. So we need strong associations between Lisa’s face and her name. Here the concept of a short-term memory rehearsal buffer in the Atkinson and Shiffrin model is quite useful. The model says that information is rehearsed in short-term memory before it is stored into long-term memory. Having information in the rehearsal buffer means that I am actively thinking about it. The longer information stays in the rehearsal buffer, the greater the probability that it will be stored in long-term memory. Not only time is important, but also the manner of rehearsal. Simple rehearsal is not very effective (remember foreign vocabulary tests in high school?). I need to do “deep” processing, which means I focus on the meaning of cue and target information. Longer and deeper processing leads to strong associations.
From this model we can take advice for every-day life.
For successful retrieval we need good retrieval cues. A good retrieval cue is strongly related to the target but not to other memories. To increase my chances to remember Lisa next time, I should focus on cues that will distinguish Lisa from others. Perhaps she has a mole or a uniquely shaped nose. I should not focus too much on things that change, such as her clothing or lipstick color. Once I have identified cues, I need to make sure they become strongly associated to the name Lisa and other things I want to remember. At the same time, I must try to ignore information that is irrelevant so that Lisa’s face will only act as a cue to memories of Lisa and not memories of cousin Sarah, who happened to be sitting nearby, telling a very animated story. Nor should I get distracted by tasty-looking snacks or my own thoughts of work that needs to be done. The best way to remember Lisa is to focus solely on her and ignore the rest of the party. Finally, during retrieval, the current context is not a good retrieval cue. I should rather think back of the previous context, last year when I spoke to Lisa. If that is not helpful, try to ignore context altogether.
If all else fails, realize that the main cause of forgetting is interference from other memories that are retrieved by your retrieval cues. The more memories you have, the more interference you will experience. Thus, forgetting is simply the result of having an interesting life.
Psychonomics Article considered in this post:
Malmberg, K. J., Raaijmakers, J. G. W., & Shiffrin, R. M. (2019). 50 years of research sparked by Atkinson and Shiffrin (1968). Memory & Cognition. DOI: 10.3758/s13421-019-00896-7.