Theravive Home

A new study published in the Journal of Nature looked at neurons, values, and our economic choices.

“The behavior we engage in, for example, when we are sitting in a restaurant and contemplating the menu,” study author Dr. Camillo Paoda-Schioppa told us. “Let’s say that there are two options – pizza or burger. How do we make that choice?”

An idea that has been around since the 18th century is that this kind of choice is made by assigning and comparing values.

“In essence, we assign a value to the pizza, a different value to the burger, and then a decision is made by comparing values,” Dr. Paoda-Schioppa told us. “The concept of value is very intuitive and frequently used, but it is also a little slippery, in the sense that it is hard – actually, impossible – to really nail it at the behavioral level.”

In other words, Dr. Paoda-Schioppa explains, by looking at choices alone, one cannot state with certainty that values are indeed computed. This issue has long been a known in economics. However, circa 15 years ago, research in neuroscience demonstrated that values are real, in the sense that neurons in the brain compute and represent the values assigned to the various options.

That result was a breakthrough, and a large number of studies subsequently confirmed the initial findings. However, it remained unclear whether and how neurons encoding values directly participate in the choice process.

“Values can guide a variety of mental processes – learning, emotion, attention, etc. – and the fact that a group of neurons represents values does not imply that these neurons are causally involved in a choice process,” Dr. Paoda-Schioppa told us. “The current study closes that gap: we show that neurons in a particular part of the brain – the orbitofrontal cortex – encode value and are causally linked to economic choices.”

Researchers knew that these neurons encoded value, and they wanted to test causality in the sense described above. Their working hypothesis was that these neurons are causally related to choices – which ended up being true.

“For almost 20 years, my research has focused on the neural mechanisms underlying economic choice behavior,” Dr. Paoda-Schioppa told us. “I find this to be a fascinating topic, with rich philosophical and societal implications. Fifteen years ago, when I was a post-doc, I found the first evidence of neurons encoding subjective values. Shortly thereafter, causality emerged as a fundamental question. However, testing causality has been difficult because of how values are represented in the orbitofrontal cortex.”

In essence, Dr. Paoda-Schioppa explains, in the restaurant setting described above, one group of neurons represents the value of pizza, and another group of neurons represents the value of the burger. To show causality, one would like to 'artificially' increase the activity of one of the two groups and, in so doing, bias choices towards the corresponding option. The artificial increase in activity can be procured through low-current electrical stimulation. The problem is that the two groups of neurons are not segregated – they are close to each other within the orbitofrontal cortex. Therefore, one cannot selectively stimulate neurons associated to one of the two options.

The experiments were conducted in monkeys who were choosing between different beverages, referred to as juices. For example, in one session, an animal might be choosing between grape juice and cranberry juice. The animal might like grape juice better; thus if offered one drop of grape juice against one drop of cranberry juice, the monkey would choose the grape. However, the animal also liked cranberry, and it was a little thirsty; thus if one drop of grape juice was offered against three, four or five drops of cranberry, the animal would eventually choose the cranberry. In other words, there was a quality-quantity trade-off.

“Our plan was to let the animal choose and, in some trials, inject a current during the offer presentation,” Dr. Paoda-Schioppa told us. “If we could show that stimulation biased choices in a predictable way, causality would be shown. To address the problem described above (different groups of neurons intermixed with each other), we played a couple of experimental tricks.”

In one experiment, researchers presented offers in sequence – first grape juice and then cranberry juice, or the other way around. Then they delivered a current at relatively high intensity during the presentation of the first offer. The effect was to disrupt the computation of value for the first offer, which biased choices in favor of the second offer.

“In another experiment, we took advantage of a phenomenon called range adaptation,” Dr. Paoda-Schioppa told us. “In essence, when we say that a group of cells represents the value of grape juice, we mean that there is a linear relation between the activity of these cells (in spikes/second) and the value of the grape. The same is true for neurons encoding the value of cranberry juice. In the experiments, the quantity of, say, grape juice offered to the monkey varies from trial to trial, within a range. For example, in one particular session, the quantity may be between 0 drops and three drops (range 0-3). Range adaptation means that the activity in response to the offer of, say, two drops of grape juice depends on the range used for that session. In particular, the activity for two drops is higher when the range is 0-3 compared to when the range is 0-6.”

When the researchers stimulate, they effectively increase the firing rate of these cells, which is equivalent to increasing the value of the grape juice. Critically, because of range adaptation, the increase in value is also proportional to the value range. The same is true for neurons encoding the value of cranberry juice. In the experiments, researchers set the two ranges (for grape and cranberry) different from each other. When they stimulated, they increased both values, but because of range adaptation, the value of the juice offered in the larger range increased more. As a consequence, the stimulation biased choices in favor of that juice. This was their prediction, and this is what they observed.

“The results confirmed our predictions and working hypothesis,” Dr. Paoda-Schioppa told us. “Experiments like this often don’t work – meaning that the stimulation has no detectable effects on behavior. So we were very pleasantly surprised. The results settle an issue that has been discussed for nearly three centuries. Going forward, there are many other interesting and open questions. For example, we don’t really know how value comparison (i.e., the decision) works. This is one of the questions we plan to pursue in future studies.”