Do you remember your first kiss? What about your grandma dying? Chances are you do, and that’s because emotional memories are at the core of our personal life story. Some rare moments are just incredibly intense and stand out from an otherwise repetitive existence of sleeping, eating and working. That said, daily life, too, is replete with experiences that have a personal emotional significance – such as disagreeing with someone or receiving a compliment.
Most of us are able to describe emotional memories in some detail, even after a long time, while memories of more mundane experiences and events fade away. But exactly why that is and how we actually remember remains unclear. In our new study, published in Psychological Review, we have come up with a computer model that may help to explain it.
To study how emotion influences memory in the laboratory, scientists typically show participants films, stories and pictures that trigger an emotional response. They may then ask the volunteers to describe what they remember. People differ greatly in their emotional responses though. Researchers therefore try to use materials that have more or less of a consistent effect on people – be it positive or negative. For example, a picture of a baby undergoing a medical procedure tends to be distressing for most of us.
Studies such as these have provided good evidence that memory really is more accurate for materials that evoke an emotional response.
Over the years, there have been a number of different ideas about why that is. One argues that people simply pay more attention to the experiences they care about – meaning they are prioritised and out-compete others. According to this theory, it’s the attention paid during the initial encoding of information that helps people more easily retrieve it later on.
But that’s not the whole story. It is clear that what happens just before and just after an experience also matters. It’s easier to recall a mildly exciting experience if it is followed by a period of quiet than if it is followed by a highly arousing event. Similarly, the particular situation in which memory is probed also influences what experiences come to mind. It is easier to recall winning a school competition when we are back at the same school for a reunion, for example.
The mathematics of memory
In our recent paper, we brought these ideas together in an attempt to provide a more coherent explanation of emotional memory. We started off by examining the information processing steps that take place in the human brain when we encode, retain and retrieve neutral information. Here we relied on an existing, established theory of memory recall which is particularly clear and precise because it expresses every one of its claims in mathematical equations.
According to this theory, each one of our experiences is linked to the mental state we have at the time – in other words, the mental context. For example, if you are in a rush one morning, then your memory of what you had for breakfast will be influenced by this wider mental context. The memory of the breakfast will also be linked to your memory of what you read in the newspaper at the same time. Such mental states change with each subsequent experience you have, but can be used later on to cue recall of past experiences. For example, if someone asks you what you had for breakfast that morning, it will help to think back to the experience of being in a hurry or reading about an accident in the news.
We then asked how emotion might modulate each one of the steps in the memory process, using findings from the experiments on emotional memory, and wrote up the potential influences in a mathematical form. Specifically, we suggested that the link between an experience and its mental context is stronger when this experience is emotional. Finally, we fed the equations into a computer program, which simulated how a person learns and recalls certain materials.
If our ideas about memory were correct, then the computer program would “recall” more accurately those items that human participants also recall better. We found that this was the case. But our model mimicked not only situations where emotion enhances memory recall, but also situations where it does not.
For example, my previous research has shown that, while people have better memory for emotional material when shown a mix of emotional and neutral pictures, this doesn’t hold up when people are shown only a series of emotional images or only a series of non-emotional images, such as someone painting a door. People may have a similar memory capacity in each such experiment. This is a bit of a mystery. But the model also produced this counterintuitive result, giving us confidence that our mathematical code may be on the right track.
Our work has a number of exciting implications. It seems that the mechanism that underlies good emotional memory is not as unique as previously thought – both emotional and neutral experiences undergo relatively similar processing. But emotion shapes the emphasis on particular steps and distinctions such as the strength of the association between items, and between items and their encoding context.
Those small changes lead to important, holistic effects on the entire process of memorising. This may be because it is so crucial for us to remember emotional experiences that evolution has shaped many aspects of remembering to be sensitive to it – such as the threat of a predator or the opportunity for food.
Because we describe the effects of emotion using mathematical equations, our work may allow scientists, one day, to predict what experiences an individual will recall. The starting point would be to try and predict which picture out of a bunch people will remember. The ultimate goal would be to try and understand this on the individual level. At the moment, there is quite a lot of uncertainty in the assumptions we make about what’s going on in any individual’s mind, particularly for how strongly different experiences are linked and how much attention they pay to experiences.
But once we gather more data on these intermediate steps, our model’s predictions may more accurately reproduce the recall pattern of individuals. Of course, we could be wrong, which would force us to revise our model. Science, after all, progresses by generating hypotheses and then testing them against empirical data.