Pavlov's dogs famously learned to associate the sound of a bell with the arrival of food, salivating in expectation of a treat whenever they heard the bell. Associative memories like these are formed when two events or stimuli are consistently paired together, allowing an animal to learn that one event predicts the other. This is useful in a predictable environment, but what happens if the two events are no longer paired? In these cases a process known as extinction enables animals to learn a new association: that the event once predicting the second event no longer serves as a predictive cue. Several lines of data strongly support the idea that extinction is new learning and not a case of unlearning. Animal studies point to two brain regions, the amygdala and prefrontal cortex, as potential key regulators of extinction learning. Amygdala activation in humans during extinction learning has been reported, but technical limitations have so far prevented characterizations of responses in human prefrontal brain regions. It is unclear whether: (1) the same brain regions mediate extinction in humans and animals; (2) associative learning and extinction learning rely on similar or distinct brain regions;and (3) a trace of the original association persists and this memory is accessible during extinction. These are the issues investigated in a recent study by Jay A Gottfried and Raymond J Dolan.
Does extinction learning in humans follow similar neurobiological principles found in animals? To answer this question the authors set out to characterize which human brain regions were active during extinction learning. While their human subjects learned associations between faces and smells, they used functional magnetic resonance imaging (fMRI) techniques to measure region-specific brain activity in their subjects. The authors showed two neutral faces, repetitively paired with two different unpleasant odours, to their human subjects and asked the subjects to sniff on every trial. Then,during extinction learning, they showed the two faces in the absence of the unpleasant odours. During extinction the authors saw enhanced neural responses in two specific brain regions, the rostromedial orbitofrontal cortex and medial amygdala. These areas are similar to those necessary for extinction learning in animals, suggesting a potential cross-species preservation of extinction mechanisms that oppose associative learning.
To test whether a trace of the original association persists during extinction, the authors decided to see if brain regions associated with the prior memory of a face-odour pair were still being activated during extinction learning. They enhanced the aversiveness of one of the unpleasant odours by increasing its intensity. When the authors presented the face that was associated with this particular odour during extinction, they noticed activation of distinct regions of the ventral prefrontal cortex, even as extinction proceeded. This implies that the subjects still retained access to the memory for the aversive odour and supports the idea that associative links are not simply erased during behavioural extinction.
Gottfried and Dolan's data show that extinction learning and associative learning in humans rely on partially independent neural systems, and that similar brain regions mediate extinction learning in humans and animals. Because of this cross-species preservation, future studies investigating extinction mechanisms in either animals or humans will no doubt complement each other.