Читать книгу The Secret Life of the Mind: How Our Brain Thinks, Feels and Decides - Mariano Sigman - Страница 29

As in the procedure sketched out by Turing, the cerebral mechanism for making decisions is built on an extremely simple principle: the brain elaborates a landscape of options and starts a winner-take-all race between them.

The brain converts the information it has gathered from the senses into votes for one option or the other. The votes pile up in the form of ionic currents accumulated in a neuron until they reach a threshold where the brain deems there is sufficient evidence. These circuits that coordinate decision-making in the brain were discovered by a group of researchers headed by William Newsome and Michael Shadlen. Their challenge was to design an experiment simple enough to be able to isolate each element of the decision and, at the same time, sophisticated enough to represent decision-making in real life.

This is how the experiment works: a cloud of dots moves on a screen. Many of the dots move in a chaotic, disorganized way. Others move coherently, in a single direction. A player (an adult, a child, a monkey and, sometimes, a computer) decides which way that cloud of dots is moving. It is the electronic version of a sailor lifting a finger to decide, in the midst of choppy waters, which way the wind is blowing. Naturally, the game becomes easier when more dots are moving in the same direction.

Monkeys played this game thousands of times, while the researchers recorded their neuronal activity as reflected by the electrical currents produced in their brains. After studying this exercise for many years, and in many variations, they revealed the three principles of Turing’s algorithm for decision-making:

(1) A group of neurons in the visual cortex receives information from the retina. The neuron’s current reflects the quantity and direction of movement in each moment, but does not accumulate a history of these observations.

(2) The sensory neurons are connected to other neurons in the parietal cortex, which amass this information over time. So the neuronal circuits of the parietal cortex codify how the predisposition towards each possible action changes over time during the course of making the decision.

(3) As information favouring one option accumulates, the parietal cortex that codifies this option increases its electrical activity. When the activity reaches a certain threshold, a circuit of neurons in structures deep in the brain – known as basal ganglia – set off the corresponding action and restart the process to make way for the next decision.

The best way to prove that the brain decides through a race in the parietal cortex is by showing that a monkey’s response can be conditioned by injecting a current into the neurons that codify evidence in favour of a certain option. Shalden and Newsome did that experiment. While one monkey was watching a cloud of dots that moved completely randomly, they used an electrode to inject an electrical current into the parietal neurons that codify movement to the right. And, despite the senses indicating that movement was tied in either direction, the monkeys always responded that they were moving to the right. This is like emulating electoral fraud, manually inserting certain votes into the ballot box.

Additionally, this series of experiments allowed for the identification of three fundamental traits of the decision-making process. What relationship is there between the clarity of the evidence and the time we take to make a decision? How are options biased by prejudices or prior knowledge? When is there enough evidence in favour of one option to call the race? The answers to these three questions are interrelated. The more incomplete the information is, the slower the accumulation of evidence will be. In the moving-dot experiment, when almost all the dots move at random, the ramp of activation in the neurons in the parietal cortex that amass the evidence is not very steep. And if the threshold of evidence needed remains the same, it will take more time to cross it; which is to say, to reach the same degree of reliability. The decision cooks over a slow flame, but eventually it will reach the same temperature.

And how is the threshold established? Or, to put it another way, how does the brain determine when enough is enough? This depends on a calculation that the brain makes in a stunningly precise way, by pondering the cost of making a mistake and the time available for the decision-making.

The brain determines that threshold in order to optimize the gains from a decision. To do so it combines neuronal circuits that codify:

(1) The value of the action.

(2) The cost of time invested.

(3) The quality of the sensory information.

(4) An endogenous urgency to respond, something that we recognize as anxiety or impatience to decide.

If, in the random-dots game, mistakes are punished severely, the players (humans or monkeys) raise the threshold, taking more time to decide and accumulating more evidence. If, on the other hand, mistakes don’t count, then the players lower that same threshold, adopting again the best strategy, which here is to respond as quickly as possible. The most notable aspect of this adaptive adjustment is that in most cases it is not conscious, and often far more optimal than we would imagine.

Consider, for example, a driver stopping at a traffic light. The driver’s brain is making a great number of estimations: the probability that the light may turn amber or red, the distance to the crossing, the speed of the car, the effectiveness of the brakes, the traffic etc. Not only this: the driver´s brain is also pondering the urgency, the consequences of an accident … In the vast majority of cases (except when something goes wrong and the monitoring system of the brain takes control) these considerations are not explicit. We are not aware of all these calculations. Yet our brains do make this sophisticated calculus, which results in a decision of when and how hard we will hit the brake pedal. This specific example reveals a general principle: decision-makers know much more than they believe they do.

In contrast with this, in some conscious deliberations (which are the only ones we do remember at the end of the day) the brain often sets a very inefficient threshold to reach a decision. We all remember having slept too long on some matters which did not require that much deliberation. For example, most of us recall deliberating ad infinitum in a restaurant between two choices even if deep inside we know we would greatly enjoy either of those two options.