The Detection of Visual Contrast in the Behaving Mouse

Abstract

The mouse is becoming a key species for research on the neural circuits of the early visual system. To relate such circuits to perception, one must measure visually guided behavior and ask how it depends on fundamental stimulus attributes such as visual contrast. Using operant conditioning, we trained mice to detect visual contrast in a two-alternative forced-choice task. After 3-4 weeks of training, mice performed hundreds of trials in each session. Numerous sessions yielded high-quality psychometric curves from which we inferred measures of contrast sensitivity. In multiple sessions, however, choices were influenced not only by contrast, but also by estimates of reward value and by irrelevant factors such as recent failures and rewards. This behavior was captured by a generalized linear model involving not only the visual responses to the current stimulus but also a bias term and history terms depending on the outcome of the previous trial. We compared the behavioral performance of the mice to predictions of a simple decoder applied to neural responses measured in primary visual cortex of awake mice during passive viewing. The decoder performed better than the animal, suggesting that mice might not use optimally the information contained in the activity of visual cortex.

Adaptable history biases in human perceptual decisions

“…Choice-history biases have been observed when subjects had to make judgments about physical weights (9), auditory stimuli (12,16,18), or visual stimuli (4,8,10,11,14,18). As in our experiment, subjects show a diversity of biases ranging from switching strategies (4,9,12,14,16), to staying strategies (10,11,18), to success-stay/fail-switch strategies (5, 6, 8, 18). These biases are typically limited to the preceding one trial and the magnitude of the choice history biases is inversely proportional to the strength of sensory stimulus, such that weaker sensory stimulus elicits stronger choice history biases (13,15,18).…”

“…Probabilistic choice model. To quantify the influence of stimulus contrast and previous trial outcome (success or failure) on current trial choices, we used a probabilistic choice model (14). The model is a binomial logistic regression which estimates the probability of selecting the right or left side based on weighting of the stimulus location (separately for each contrast), success and failure outcome on the previous trial and overall bias (preference of one side over the other).…”

“…However, under such conditions, human observers exhibit nonsensory biases (9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20) For instance, in the classical two-alternative forced choice task, a subject is asked to make a perceptual decision based solely on the present sensory evidence (21) and to use success or failure feedback only to optimize their use of such sensory information (22). However, both mice (14) and humans (15,17,18) make choices that are biased by their recent history of successes and failures and are thus suboptimal. Such biases, where prior beliefs rather than evidence guide behavior, are a form of fallacy, similar to the gambler's fallacy or hot-hand fallacy (23, cf.…”

“…For example, in one of the conditions, if the subject failed on the previous trial by choosing left, the stimulus would be presented on the right with 80% probability. We used a probabilistic choice model (14) to document each subject's choice history biases, such as preference to stay on the same side after a failure or switch after a success. By measuring the biases, this model allowed us to assess their adaptability, by comparing the conditions where trial statistics were manipulated to the baseline condition where they were truly random.…”

Multiplexed modulation of behavioral choice

“…In the absence of additional sensory input, each alternative behavior will occur at chance levels. This phenomenon has been observed in many versions of the twoalterative forced-choice task (Fechner, 1889) -in a variety of taxa such as birds (Klump et al, 1986;Nelson and Suthers, 2004), mice (Busse et al, 2011), monkeys (Britten et al, 1993) and humans (Green and Swets, 1966) -where, in absence of compelling input, the likelihood of each behavior hovers at chance. Under our model's assumption, leeches have sensors that are activated by shallow water, but have no sensors that are activated by deep water.…”