Like many animals, we thrive to repeat the behaviors that have paid off in the past and avoid those that did not. This principle is at the heart of most learning theories. It is also one of the most important functions of the brain: to adjust behaviors according to our experiences.
However, we seem to also infer the value of things we have never experienced. I have never driven an antique 1923 Overland Model 91 and I assume most of you never did either. Yet, imagine we were offered a choice between driving this car for one hour or obtaining twenty dollars. Any of us could come up with decision that would reflect our personal preferences. This is because we can infer the value of driving that car based on our general knowledge, before having actually tried it.
In the last decades, scientists have found a set of brain areas in humans as well as in other animals that seem to be triggered by the value of options in the environment. The orbitofrontal cortex – the part of the brain located just behind our eyes – is one of those areas that can be activated by rewards such as money, juice or even viewing photos of other individuals1,2,3. Whether this part of the brain plays a role in caching the value of what was already experienced or inferring the value of novel events or contexts that were not yet experienced had up to now remained unknown.
Recently, a team of scientists led by Geoffrey Schoenbaum has published a discovery in Science suggesting that the orbitofrontal cortex plays a role in learning inferred value but not in situations when cached value is sufficient4. They used a clever experimental setup in which rats first learn an association between two sounds (i.e. when Sound #1 comes up, Sound #2 will come up right after). The rats are then exposed to sessions in which Sound #2 is played along with the delivery of a reward in a food cup. Thus the rats learn through direct experience that Sound #2 is associated with a reward. This learning can be done by caching past experiences. When this learning is completed, they then tested how rats react to Sound #1. Because Sound #1 was never paired with a reward, the rats are not expected to attribute value to it, unless they make an inference: that Sound #1 is followed by Sound #2, which is followed by a reward. This learning is thus inference-based and rats appear to be able to do it as they display interest in the food cup when the first sound is played, despite the fact that it was never linked to any reward.
Interestingly, the authors of the study found that blocking brain activity temporarily in the orbitofrontal cortex impairs inference-based learning, but not learning based on caching past experience. The authors propose a quite ambitious speculation – that the role of the orbitofrontal cortex might not simply be to cache the value of options that are available to animals but that it may play a role in manipulating known associations about the environment in a model-based fashion (i.e. predicting what will happen in the environment by modelling its different components and the interactions between them). This exciting study makes an important step forward in understanding the specific role of the orbitofrontal cortex in the way animals learn about their environment and interact with it.
1. Watson KK, Platt ML. (2012) Social Signals in Primate Orbitofrontal Cortex. Current Biology.
2. Padoa-Schioppa C, Assad JA. (2006) Neurons in the orbitofrontal cortex encode economic value. Nature.
3. Levy DJ, Glimcher PW. (2012) The root of all value: a neural common currency for choice. Current Opinion in Neurobiology.
4. Jones et al., (2012) Orbitofrontal cortex supports behavior and learning using inferred but not cached values. Science.