Betting on Awareness

Conscious awareness is difficult to measure. On the one hand, it's a private, subjective phenomenon that resists easy quantification. (Only I know what I am aware of.) On the other hand, neuroscience won't be able to understand many mental phenomena - like consciousness - unless it can develop objective measurements for these slippery, subjective phenomena.

Traditional measurements of awareness have relied on self-reports of confidence. For example, experimental subjects might be asked to judge whether a cloud of dots moved to the left or to the right. They are then asked to report how confident they are by entering a number on a scale of 1 to 5, with 5 being the most confident. The assumption is that more confidence signifies greater awareness. Alas, this experimental method has proven to be very unreliable. Different people have different definitions of "confidence," and small changes in the experimental instructions can have large ramifications in the experimental data.

So what should neuroscience do? Is awareness impossible to measure? In the latest Nature Neuroscience, Navindra Persaud, et. al. have hit up a brilliant new method: have subjects bet on their confidence. Rather than asking subjects to explicitly report on their levels of confidence, they are asked to wager some money on each decision. The assumption here is that people will bet larger amounts of money when they are confident of knowing the answer. In other words, levels of awareness will be approximated by the levels of the bets.

In this Nature Neuroscience paper, Persaud et. al. use three different paradigms to test out this wagering protocol: visual detection task in a blindsight subject, and an artificial grammar task and the Iowa gambling task in normal subjects. I'll just focus on the visual detection task in the blindsight patient, although the two other paradigms are also worth reviewing.

Blindsight is a profoundly weird phenomenon. It happens when patients suffer lesions in their V1. Although these patients think they are blind, they can actually see, at least unconsciously. What they are missing is awareness. While their eyes continue to transmit visual information, and undamaged parts of their brain continue to process this information, blindsight patients are unable to consciously access what their brain knows.

So how can you tell blindsight and blindness apart? Blindsight patients exhibit an astonishing talent. On various visual tasks they perform with an aptitude impossible for the totally blind. For example, they can "guess" with uncanny accuracy whether they were shown a square or a circle, or whether a light has been flashed. While they have no explicit awareness of the light, they can still respond to it, albeit without knowing what they are responding to it.

In this Nature Neuroscience paper, Persaud et. al. use their innovative protocol to confirm these facts about blindsight. Their blindsight patient "guessed" correctly 70 percent of the time, which is far above random chance. Yet he failed to make money of his superior performance: only 48 percent of his correct choices were associated with a high wager. Furthermore, 39 percent of his false choices were also associated with a high wager. Because he lacked conscious awareness of the stimuli, he was unable to convert his sensation into winnings. But this wasn't because he didn't know how to bet. In a control experiment, the authors demonstrated that when the blindsight patient is aware of the stimulus, he wagers high on almost every trial, which is what normal people do. As Kristof Koch notes in his review of the article, "His wagering thus seemed to mirror his awareness rather than his performance on the task, suggesting that wagering may provide a means to measure awareness."