Researchers Begin to Decode Decision-making Processes

by Tom Valeo

December 10, 2008

A decision follows what may feel like conscious deliberation, but research suggests that our choices take shape below the threshold of consciousness, with the brain rapidly integrating sensory input, memory and the probability of reward.

Moreover, this decision-making machinery is easily disrupted by drugs, sleep deprivation and damage to brain regions essential to the process.

At the Society for Neuroscience meeting, Michael Shadlen, an investigator with the Howard Hughes Medical Institute and a professor at the University of Washington Medical School, compared the decision-making process to the ingenious method British mathematician Alan Turing devised to crack the German Navy’s Enigma code during World War II. The Enigma machine, which resembled a typewriter, contained multiple rotors that replaced each letter in the original message with a substitute. Because the rotors were changed constantly, each letter of an intercepted message confronted decoders with a multitude of possibilities.

A key step that helped Turing break the code involved the creation of a mathematical formula that assigned a “weight” to evidence supporting the possibility that letters from different messages came from machines using the same settings. Once the weight of evidence passed a certain threshold, Turing made the decision that the machines did indeed share the same settings.

“The mathematical principles Turing developed to break Enigma are similar to the ones our own brain uses,” Shadlen said.

To demonstrate, Shadlen inserted microelectrodes into the parietal lobe of a monkey trained to watch a computer screen and decide if flittering dots were drifting to the right or the left. A correct guess earned the monkey a sip of juice.

The neurons in the parietal lobe, located at the crown of the head, accumulated evidence coming from the monkey’s visual cortex as it stared at the dots. Like the threshold for evidence that Turing established for decoding Enigma messages, when the evidence appeared strong to the monkey, the parietal neurons emitted a strong signal that built up quickly, leading to faster and more accurate choices. By monitoring the signals coming from the region, researchers could predict which choice the monkey would make.

“We all engage in such probabilistic reasoning, which is what Turing himself suspected,” Shadlen said. “We accumulate evidence, and when we reach a satisfactory threshold of probability, we decide.”

In an experiment not yet published, Shadlen monitored the monkey’s confidence in its decisions by offering a small but guaranteed reward if, instead of looking to the right or left to indicate which direction the dots were drifting, it looked at a third target, thereby indicating, in effect, “I’m not sure.” This experiment exposed what Shadlen calls the “degree of certainty” the monkey had in its decision making.

Related experiments with humans have persuaded him that we engage in a similar process when we reverse a decision. “There is often additional information wending its way through the brain,” he said.

The response to mistakes

But what happens when we make a wrong decision?

The orbitofrontal cortex (OFC), located at the front of the brain behind the eyes, reacts to the mistake and thereby helps us alter our behavior, according to Geoffrey Schoenbaum, an assistant professor in the department of anatomy and neurobiology at the University of Maryland School of Medicine.

“The OFC enables us to recognize when things don’t go as we expected,” Schoenbaum said. “Damage to the OFC results in deficits in what we call reversal learning—the ability to go back and reconsider a decision based on the outcome. The OFC is well-known for its role in good judgment, for allowing us to use what we think we know about likely outcomes to guide decision making, but the OFC also enables us to recognize and learn from things that don’t go as we expected.”

However, under the influence of cocaine and other drugs, the OFC falters, undermining an addict’s efforts to quit taking drugs in light of negative consequences. Schoenbaum believes cocaine may actually damage the OFC.

“Changes in the OFC may be involved in the addict’s compulsive seeking of drugs,” he said. “There’s some change in the brain that prevents modulation of drug-taking behavior.”

Like drugs, lack of sleep compromises decision making. Vinod Venkatraman and his colleagues at Duke University and National University of Singapore demonstrated that sleep-deprived gamblers are more inclined to make high-risk bets, apparently because they overestimate their chances of winning.

The researchers scanned the brains of 28 young adults (with a mean age of 22.3 years) who were playing a gambling game that involved various risk strategies. They repeated the scans after the subjects had gone for 24 hours without sleep. The participants no longer embraced a strategy of avoiding losses; instead they went for big gains despite the threat of big losses.

Brain scans showed that when the subjects were sleep-deprived, they displayed less activation in the insular cortex, a region associated with negative moods and emotions. Gains produced greater activation in the striatum, which contains many receptors for dopamine, the neurotransmitter released abundantly after taking cocaine or other recreational drugs.

“Sleep deprivation promotes risk-seeking behavior by increasing neural responses to anticipated gains in regions associated with reward processes, particularly the ventral striatum and ventral medial prefrontal cortex,” Venkatraman said. “It also diminishes sensitivity to losses.”

Missing raisin sparks brain activity

The habenula, a small structure located deep within the brain at the rear of the hypothalamus, also reacts to negative expectation or disappointment. Okihide Hikosaka and a colleague, both of the National Institutes of Health, stumbled upon this function while recording readings from the brain of a monkey trained to follow a spot displayed on a computer screen.

The monkey received a raisin each time it performed correctly, but one day the two researchers decided to be “slightly mischievous,” Hikosaka said, by offering a raisin with a hand that turned out to be empty. “Gazing at the empty hand, the monkey raised his eyebrows and showed his teeth, disgruntled to be denied his tasty treat,” Hikosaka recounted in a written summary of work he presented at the meeting. “At the same time, many neurons in the habenula emitted a burst of activity in unison, as if it had been the expression of anger or disappointment.”

Further testing revealed that the lateral habenula reacts strongly when expected rewards are denied or replaced by mild punishment, such as a puff of air to the face. Dopamine neurons are inhibited by the habenula’s activity, and because dopamine contributes to learning by producing a positive sensation in response to success, the habenula contributes to learning by shutting off dopamine in response to disappointment.

“Habenula neurons and dopamine neurons appear to have a reward and punishment relationship,” Hikosaka wrote. “Some previous studies have indicated that excessive habenular activation is correlated with depressive mode and behaviors. Furthermore, there have been suggestions that psychiatric disorders, including depression and schizophrenia, are caused by abnormal activity of habenular neurons.”

Dopamine produces pleasant sensations that influence our decisions. Eating a chocolate cookie may produce a surge of dopamine, for example, and promote the decision to reach for another. Spending time with friends or solving a puzzle may have a similar effect.

But what shapes moral decisions?

Unlike Immanuel Kant, who believed that emotions should be banished from moral decisions, philosopher Patricia Churchland believes that neuroscience points to the opposite conclusion—that emotions play a crucial role in moral choices.

Moral behavior is rooted in social relationships, according to Churchland, and those relationships depend on feelings of attachment and trust. Oxytocin and vasopressin, two neurochemicals in mammals, foster such feelings and appear to be the source of many animals’ willingness to sacrifice personal well-being for the benefit of the group.

One striking example of oxytocin’s effect occurs in the prairie vole and the montane vole. Prairie voles mate for life, with the male taking an active part in caring for offspring and protecting the nest. Montane voles, in contrast, do not show preference for one mate and display little social activity. The females raise offspring alone.

Each type of vole has oxytocin receptors, but prairie voles have a greater density of oxytocin receptors in the nucleus accumbens, a structure at the front of the brain that contributes to the sensation of pleasure. Prairie voles also have many receptors for vasopressin in the ventral pallidum, a deep brain structure that is also part of the brain’s reward circuit.

Many scientists now believe that the action of oxytocin and vasopressin on the brain are crucial to the conspicuous sociability of prairie voles, and Churchland pointed out that humans, another strongly social species, also show greater trusting behavior in response to increased levels of oxytocin.

Churchland believes that altruism and group loyalty among humans depend on feelings of attachment and trust, which depend on oxytocin and vasopressin. People may feel as though they have chosen to be loyal to family and friends after deliberating about whether that’s the right thing to do, but such actions undoubtedly spring, at least in part, from deep neurobiological sources.

 “Attachments play a large role in how we conduct ourselves in the social domain, but that doesn’t mean we have no control or choice over our behavior,” Churchland said. “However, it does mean that, contrary to what the existentialists argued, we do not create all of our behavior after conscious reflection.”