We are making decisions based on our sensation in our daily life. When playing football you may want to throw a ball to the place where fewer opposing players are running. However, sometimes you have to make decisions without full confidence, for example when you can only glance at a group of players running around for a very short time. In addition to the decision itself (the place you believe has least opposing players), your degree of confidence may also influence your behavior. Realizing lack of confidence, you may decide to run instead. How we make decisions from sensory inputs, and how we deal with degree of confidence is an important problem in neuroscience. The laboratory of Dr. Michael Shadlen at Columbia University (recently moved from University of Washington) has done pioneering works in this field.
In their recent paper, Kiani and Shadlen (2009) reported neural activities closely related to behavior governed by degree of confidence. They used random-dot visual stimuli composed of a proportion of dots moving coherently in one direction and the other dots moving around randomly. After watching this stimulus for a while, monkeys are asked to report whether the direction of the stimulus was toward one target or the other by a looking at one of the targets. When most of the dots are moving coherently, the direction is obvious, but when most of the dots are moving randomly, it is hard to tell the direction.
Motion information of the stimulus is transferred to the sensorimotor area called the lateral intraparietal area (LIP) from the visual cortex. LIP controls saccadic eye movements, and the neurons in LIP have a receptive field representing a target area of saccades. It has been shown that the neurons in LIP ramp up their activity when the random dot stimulus is moving toward the neurons’ receptive field and decrease when it is in the opposite direction. Rate of the increase is proportional to the motion coherence of the stimuli, and if monkeys can make saccades anytime they want, they do so when the firing rate reaches a certain value, as if there are thresholds to be confident to make a decision. Then, does intermediate firing rate mean lack of confidence?
They answered this question by showing another “sure” target in half of the trials (See Figure 1). The monkeys can get a certain but small reward if they waive decisions and look at this sure target. First, the monkeys performed better when they choose not to waive decisions than when they were forced to make decisions in the trials without the sure target. This shows that they can utilize their degree of confidence to maximize rewards. Second, the monkeys waived decisions more often when the firing rate of a LIP neuron is intermediate. These results show that the activities in LIP reflects the degree of confidence rather than just representing where to saccade next and that these activities are relevant to the behavior of waiving decisions.
In another paper, Yang and Shadlen (2007) showed that the activities in LIP corresponds to how likely the target lead to reward even when the stimuli were not moving random-dots, but some symbols probabilistically associated with either choice. LIP is not only accumulating stimulus motion direction, but represents more abstract thing, namely probabilistic belief in the target leading to a reward.
To find out more recent work from his lab, please join us on Tuesday, February 5, 2013, at 4:00pm in the Large Conference Room of the Center for Neural Circuits & Behavior for Dr. Shadlen’s talk, “Believing and time: A neural mechanism for decision making.”
Akinori Mitani is a first-year neuroscience Ph. D student currently rotating in Massimo Scanziani’s Laboratory.
Kiani R. & Shadlen M.N. (2009). Representation of Confidence Associated with a Decision by Neurons in the Parietal Cortex, Science, 324 (5928) 759-764. DOI: 10.1126/science.1169405
Yang T. & Shadlen M.N. (2007). Probabilistic reasoning by neurons, Nature, 447 (7148) 1075-1080. DOI: 10.1038/nature05852