Changes in patterns of brain activity predict fear memory formation
Brain activity predicts whether a fearful event will become a lasting fear memory, reports a study recently published online in ‘Nature Neuroscience’. Using a new method for analysing functional magnetic resonance imaging (fMRI) data the authors showed that the long-term storage of fear memory is preceded by changes in patterns of brain activity. This study may provide a basis for understanding how fear memories are strengthened and kept – a process that may go wrong in anxiety conditions such as post-traumatic stress disorder.
Authors Renée Visser MSc., Dr Steven Scholte, Tinka Beemsterboer MSc. and Prof. Merel Kindt (all researchers at the University of Amsterdam) report that they can predict future fear memories by looking at patterns of brain activity during fearful experiences. During magnetic resonance brain imaging, participants saw pictures of faces and houses, some of which were followed by a small electric shock. Weeks later, people returned to the lab and the authors measured their fear responses to the pictures they had previously seen. The authors report that this later fear memory was associated with more similarity in the patterns of brain activity when people looked at the pictures that were paired with shocks. Those who had strong fear responses showed a greater similarity in their responses to the different images associated with shocks during the initial scanning session than those who had weak fear responses.
Method for analysing data
The authors used a method for analysing fMRI data that differs from standard analytical techniques in two ways. First, they looked at the spatial pattern of activation (multi-voxel pattern analysis, or MVPA), instead of looking at the average activation in a certain brain area. The spatial pattern of activation is different for each stimulus presentation and therefore a unique representation of that stimulus at that particular moment. Second, single trials were not collapsed, but analyzed separately. By correlating the spatial patterns related to consecutive presentations of the same picture the authors were able to assess how the representation of a stimulus changes as a function of fear conditioning. They showed that pictures that were paired with a shock (and thus became predictors of threat) elicited more refined patterns of activation than pictures that were never followed by an aversive outcome. Furthermore, it appeared that the neural representations of pictures that were initially unrelated (faces and houses) became more alike when they predicted a shock, but not when they did not predict threat. Threatening stimuli thus became associated at a ‘higher’ level. The degree to which this occurred indicated the formation of a fear memory: the stronger the higher-order association, the stronger the long-term expression of fear.
New ways of predicting and examining fear memories
These findings are remarkable given that until now no markers were available that predicted whether fear learning would develop into fear memory. It was not even clear whether the selection of information for storage into long-term memory takes place during learning, or shortly after. The expression of fear during learning, as for example indexed by subjective measures and pupil dilation responses, is not a predictor of the long-term expression of fear. Virtually everyone shows a fear response while experiencing a threatening event, but not everyone stores this information into long-term memory. Changes in average brain activation, the measure of interest in most fMRI studies, also do not tell anything about fear responses a few weeks later. This study shows that by looking at patterns of activation it is actually possible to predict long-term fear memory.
This study may provide a basis for examining the conditions under which fear memories are strengthened, maintained or even erased.
Visser, R. M., Scholte, H. S., Beemsterboer, T. & Kindt, M.: ‘Neural pattern similarity Predicts Long-term fear memory' in Nature Neuroscience (2013).