How Do We Make Decisions?

 

[   C  H  R  O  N  I  C  L  E   ]

 

In the fog, a driver wonders whether he should turn left or right. He thinks he sees the house he is looking for, the wood looks familiar, the gate is indeed the one he remembers, and the thought strikes him; he’s sure, he needs to turn right. How did this decision-making process occur? A well-established fact in psychology is the close link between the uncertainty of some information we have (very high uncertainty in the fog), and the time spent making a decision. Psychologists have tested this cognitive process in many forms.

One possible experiment consists of displaying a cloud of one hundred points moving in random directions, on a screen. If only two of these points follow parallel trajectories and in the same direction (we speak of 2% coherence), the overall movement still appears random to the observer. However, if fifty points are moving like this (50% coherence), the sensation that this set of points is moving in the same direction is immediate. The lower the coherence, the more time the observer takes to indicate the overall movement, and hence to decide. What exactly is happening in the brain?

At the start of the 2000s, the American scientists Michael Shadlen (Columbia University) and William Newsome (Stanford University), studied this issue in monkeys. They recorded the electrical activity of neurons in the lateral intraparietal cortex, a region of the brain associated with eye movement control. The primates were trained to indicate the direction of movement of a set of points by directing their gaze to one side or the other (1). Result: neuron activity increased gradually during the observation, until it reached a peak when the animal made a choice. These neurons therefore gradually accumulate sensory clues until they reach a level of confidence, triggering the decision. This is a little like when a bottle is being filled under a tap whose jet is more or less dispersed; once the bottle is full, the choice is made, which can take some time if the jet is not very “coherent” and if the water flows more outside than inside the bottle.

This remarkable discovery corroborates numerous mathematical models that predict the existence of a gradual accumulation signal in the brain. However, one US study published by Jonathan Pillow and his students at Princeton and Texas University cast doubt on this mechanism, pointing out that Michael Shadlen and William Newsome calculated the average electrical activity of the neurons over a large number of tests. Jonathan Pillow’s team reanalysed the data, this time test by test (2). Using elaborate mathematical tools, the team showed that when the point cloud was presented, the neurons moved from an “unexcited” state to an “excited” state without transition.

Now there is a hot debate between the two teams, who are battling it out in comments in specialist journals. Each one is sticking to its original position. A decision needs to be made however, as understanding irrationality in our choices is a major issue in neuroeconomics. If we want to understand the biaises that affect our decision-making, the underlying neuronal processes needs to be decoded.

New experiments are therefore needed, in particular to record the electrical activity of several neurons simultaneously. This would confirm if the activity observed is indeed linked to a network effect, instantly switching all neurons from one state to the other. This would mean that the neurons were incapable of indicating anything other than “I don’t know” or “I know”. In this case, where does the accumulation of clues explaining so neatly the time needed to make a decision, come in? Perhaps in the synapses - microscopic contacts used by the neurons to communicate with each other...

 

(1) M. N. Shadlen and W. T. Newsome, J. Neurophysiol., 86, 1916, 2001.

(2) K. W. Latimer et al., Science, 349, 184, 2015.

 

> AUTHOR

 

Adrien Peyrache

Neuroscientist

Adrien heads at McGill University in Canada a research laboratory devoted to studying the neuronal processes involved in spatial navigation and memory.

Share on Facebook
Share on Twitter
Please reload

RECENT POST

November 11, 2018

November 11, 2018

Please reload