Welcome to the seventh installment of

Mind Matters

Mind Matters is Sciam.com's "seminar blog" on the sciences of mind and brain. Each week, top scientists discuss the research driving neuroscience, psychology, and psychiatry. You can join them. We hope you will.

This week we look at

Betting on Consciousness

What's consciousness? Perhaps, a new study suggests, it's what you're willing to bet on. Photo by Adrian Sampson, Creative Commons license.


by David Dobbs, Editor, Mind Matters

Consciousness -- the awareness of perception, sensation or thought -- presents one of neuroscience's most slippery problems. How do we become aware of something? When can we be said to be aware of something? Being tired and being conscious of being tired are two different things; how does one become the other? Philosophers, having debated these heady questions for centuries, have recently been joined by neuroscientists determined to test them with replicable measurements of awareness. In the paper described here -- "Post-decision wagering objectively measures awareness," published in Nature Neuroscience on 21 January 2007 -- the authors, Navindra Persaud of the University of Toronto and Peter McLeod and Alan Cowey of Oxford University, made a splash bringing to these ancient questions an ancient measure of confidence: gambling. In a set of experiments drawing equally on Oliver Sacksian neurological anomalies and Friday night poker, Persaud and colleagues manage to test awareness of knowledge that the subjects don't even know they have. Asking subjects to bet on how well they know something is an established trick in neuroeconomics. But as our experts Christof Koch and Kerstin Preuschoff write, applying this tactic to awareness appears to give consciousness researchers a much-needed but elusive tool -- a way to assess consciousness without disturbing it.

Measuring Consciousness: A Neuroeconomic Gamble Pays Off

by Christof Koch and Kerstin Preuschoff
Much of what we do goes on outside the pale of consciousness -- whether we adjust our body posture or decide to marry someone, we often have no idea why or how we do the things we do. The Freudian notion that most of our mental life is unconscious is difficult to establish rigorously. While it seems easy to answer the question "Did you (consciously) see the light turn on?", more than a hundred years of research has shown otherwise. The key problem is defining consciousness such that it can be measured independent of the internal state of an individual's brain while still capturing its subjective character. One common experimental assessment of consciousness -- or awareness of sensation, perception or thought -- is based on "confidence." For instance, a subject has to judge whether a cloud of dots on a computer screen moves to the left or to the right. She then reports how confident she is by a number -- for example, 1 to indicate pure guessing, 2 for some hesitation, and 3 for complete certainty. This procedure assumes that when the subject has little awareness of the dots' direction of motion her confidence is low, whereas if she clearly "saw" the motion, her confidence is high. In "Post-decision wagering objectively measures awareness," Persaud, McLeod and Cowey introduce a more objective measure of consciousness that exploits people's desire to make money. This method is adapted from economics, where it is used to probe a subject's belief about an event's likely outcome: people who know that they have information are willing to bet on it, that is, they are willing to put their money where their mouth is. Think of investing in mutual funds: the more certain you are that high technology will do well over the next year the more money you will allocate to a technology sector fund. Persaud and colleagues use this sort of wagering to reveal consciousness or lack thereof. In the experiments reported here, subjects do not directly state their confidence in their awareness. Instead, subjects first make a decision regarding whether they've perceived something and then must wager either a small or a large amount of money on their confidence in this decision. If the subject's decision proves correct, she wins this money; otherwise, she loses it. The optimal strategy is to bet high whenever she feels that she is not just guessing. The experimentalists apply this wagering technique to three examples of nonconscious processing. The first experiment involves patient GY, who has "blindsight" due to a car accident that damaged parts of his visual brain. This condition leaves him with the ability to locate a light or report the direction in which a bar is moving while denying having any visual experience; he insists that he is simply guessing. GY does not disappoint in this study: when asked to indicate the presence or absence of a faint, small grating on a computer screen, he does so correctly in 70 percent of all trials, far above chance (50 percent). Yet he fails to convert this superior performance into money when wagering; he places a high bet on only about half (48 percent) of his correct choices. When GY is consciously aware of the stimulus, he wagers high, much as you or I would. His wagering thus seems to mirror his conscious awareness of the stimulus (that is, his belief that he saw it) rather than his actual (unconscious) detection of the stimulus, suggesting that wagering may provide a means to measure awareness. The second experiment involves an artificial grammar task. In this task participants learn a small number of short letter strings. They are then told that the strings obeyed a simple rule (of the sort, for example, that perhaps every "x" is always followed by an "a"). But they are not told what the rule is. When shown a new string, subjects can more often than not correctly determine whether the new string follows the unknown rule. Yet only rarely can subjects articulate why they believe a string does or does not obey the rule. The overall rate of correct classification (81 percent) is far better than chance (50 percent). Yet subjects do not convert performance into money. High wagers follow a correct choice 45 percent of the time and follow a false choice 32 percent of the time. In short, the subjects are usually right about whether the string follows the rule, but they lack enough confidence to bet on it. In the final experiment, the Iowa gambling task, subjects pick the top card from one of four decks. Each card wins or loses the subject a certain amount of money. Unbeknownst to subjects, two of the four decks have a net positive yield and two a negative yield. Subjects must place a low or a high wager on the chosen card before it is revealed and lose or win accordingly. In the test, the subjects turn scores of cards over, one at a time, each time finding out whether they win or lose. They almost always figure out which decks are winners and start to pull cards mostly from those decks -- but they usually turn over at least 30 cards on those decks before they gain the confidence to bet aggressively on the results. That is, subjects only start to make money long after their own behavior should have revealed that they knew which decks were winners. To explore this hesitance, Persaud and colleagues used a second variant of this experiment in which they queried the subjects every tenth trial about everything they knew about the game and the decks. When the subjects thus examined their knowledge about the game, the gap between the onset of positive deck selection and onset of advantageous betting disappeared -- suggesting that the act of introspection alters subjects' awareness. Examining their knowledge made them more aware of what they knew. This demonstrates that if subjects learn to trust their gut instincts -- betting on knowledge they are not yet aware of -- they can do better, a demonstration of the utility of the leitmotif of Western philosophy, "Know thyself." The wagering techniques used by Persaud, McLeod and Cowey rely on people's instinct for reaping a profit. Compared to forcing subjects to become aware of their own consciousness -- and in the process perturbing the very phenomenon one wishes to measure -- wagering provides a more subtle way to assess awareness. This is an exciting and revealing new way to study awareness and consciousness. From such small steps comes progress in answering the age-old question of how consciousness arises from experience.
Christof Koch is a professor of biology and engineering at the California Institute of Technology, where he investigates visual awareness and how the brain produces consciousness -- a venture he describes in his book The Quest for Consciousness.

Kerstin Preuschoff is a postdoctoral scholar in Decision Theory and Neuroscience at the California Institute of Technology, where she investigates decision-making under uncertainty in the human brain and uses brain imaging to explore neural activity during gambling tasks.