Sources of noise during accumulation of evidence in unrestrained and voluntarily head-restrained rats

  1. Benjamin B Scott
  2. Christine M Constantinople
  3. Jeffrey C Erlich
  4. David W Tank  Is a corresponding author
  5. Carlos D Brody  Is a corresponding author
  1. Princeton University, United States
  2. New York University Shanghai, China
7 figures

Figures

Figure 1 with 2 supplements
Visual accumulation of evidence task, unrestrained version.

(A) Task schematic. A rat initiates a behavioral trial by inserting his nose into a center port along one wall of an operant training box (left panel). The rat is presented with a series of …

https://doi.org/10.7554/eLife.11308.003
Figure 1—figure supplement 1
Stimuli and performance.

(A) A histogram of the number of left flashes presented on all trials included in this paper (data is pooled across all unrestrained rats). (B) Histogram of the number of right flashes presented on …

https://doi.org/10.7554/eLife.11308.004
Figure 1—figure supplement 2
Psychometric performance of individual rats on the visual accumulation of evidence task.

Behavioral performance of each rat included in this study, including trials with up to 10 flashes on each side. Rats are sorted by number of behavioral trials included in each plot.

https://doi.org/10.7554/eLife.11308.005
Figure 2 with 4 supplements
Error rate increases with number of flashes, not trial duration.

(A) For trials of fixed duration and fixed difference in number of flashes (Δ-flashes, colored lines), behavioral performance decreased with more flashes. (B) Reverse correlation analysis indicating …

https://doi.org/10.7554/eLife.11308.006
Figure 2—figure supplement 1
Fits of drift diffusion-like model to individual rats.

(A) Schematic of the accumulation model of Brunton et al. (2013) used here to compare the contribution of flash- and time-associated noise to behavioral variability. At each moment in time the model …

https://doi.org/10.7554/eLife.11308.007
Figure 2—figure supplement 2
Effect of flash number on performance of individual rats.

Changes in behavioral performance (% correct) for each rat as a function of the number of total flashes presented. Data points indicate behavioral performance relative to the average performance (Δ …

https://doi.org/10.7554/eLife.11308.008
Figure 2—figure supplement 3
Effect of trial duration on performance of individual rats.

Changes in behavioral performance (% correct) for each rat as a function of trial duration. Δ Performance was computed by estimating the average performance (% correct) across all trials with a …

https://doi.org/10.7554/eLife.11308.009
Figure 2—figure supplement 4
Psychophysical reverse correlation for each individual rat.

Reverse correlation analysis indicating the relative contribution of flashes occurring at different times in the trial to the subject’s behavioral choice. Each point on the upper (red) lines …

https://doi.org/10.7554/eLife.11308.010
Figure 3 with 7 supplements
Signal detection theory-based model reveals linear scaling of standard deviation of numerical estimates.

(A) Schematic of the model used to determine the standard deviation (σ) of the subjects’ estimate of flash number. Left panel indicates the stimuli from an example trial in which four flashes were …

https://doi.org/10.7554/eLife.11308.011
Figure 3—figure supplement 1
Signal detection theory-based model fit to individual rats.

Results of fitting the model described in Figure 3 for each rat: model fits of the standard deviations (σ0σ15) in each rat’s estimate for different numbers of flashes. Error bars indicate the 95% …

https://doi.org/10.7554/eLife.11308.012
Figure 3—figure supplement 2
SDT model prediction vs. data for each rat.

Comparison of psychometric performance of each rat (data, blue circles) and the signal detection theory-based model prediction of performance (model, green lines). For each rat, the model …

https://doi.org/10.7554/eLife.11308.013
Figure 3—figure supplement 3
Behavior and SDT model approximate scalar variability.

Scalar variability (and Weber’s law) predicts that scaling the number of flashes by the same factor would lead to identical discriminability/performance. Here we sought to test that directly. …

https://doi.org/10.7554/eLife.11308.014
Figure 3—figure supplement 4
Signal detection theory-based model fit to the auditory (clicks) data.

Each panel shows model fits of the standard deviations (σ) in the rat’s estimate of the effective number of clicks for an individual rat performing the auditory accumulation of evidence task. Error …

https://doi.org/10.7554/eLife.11308.015
Figure 3—figure supplement 5
Permutation test comparing goodness-of-fit of scalar variability and linear variance to the auditory (clicks) data.

A nonparametric permutation procedure (see Materials and methods: Model Comparison) was used to evaluate whether the MLE fits of the standard deviations in the rat’s estimate of number of effective …

https://doi.org/10.7554/eLife.11308.016
Figure 3—figure supplement 6
Chronometric plots for the auditory (clicks) data and predictions of scalar variability.

Each panel shows the performance of an individual rat on the auditory accumulation task as a function of the duration of the cue period. Colored dots indicate the data and colored lines indicate the …

https://doi.org/10.7554/eLife.11308.017
Figure 3—figure supplement 7
Chronometric plots for the auditory (clicks) data and predictions of scalar variability with an offset.

Each panel shows the performance of an individual rat on the auditory accumulation task as a function of the duration of the cue period. Colored dots indicate the data and colored lines indicate the …

https://doi.org/10.7554/eLife.11308.018
Figure 4 with 2 supplements
Comparison of behavioral models suggests the presence of at least two accumulators.

(A) General forms of the signal detection theory-based models that were compared assumed either single or dual accumulators. Each model determines which choice to make on each trial, by randomly …

https://doi.org/10.7554/eLife.11308.019
Figure 4—figure supplement 1
Model comparisons for each rat.

Bar plots indicate the difference in the likelihoods between the scalar variability, dual accumulator static sampling model (b) and the seven other model versions, described in Figure 4, for each …

https://doi.org/10.7554/eLife.11308.020
Figure 4—figure supplement 2
Model predictions versus data.

Performance of each model described in Figure 4. Red circles are data pooled from all rats. Green lines are model predictions, using parameters fit to the pooled data. The 'Scalar variability dual …

https://doi.org/10.7554/eLife.11308.021
Rats accumulate flash number, not duration.

(A) Distribution of flash durations. In a subset of experiments, flash durations were drawn from a Gaussian distribution with a mean of 10 ms. As a consequence, on some trials in which the same …

https://doi.org/10.7554/eLife.11308.022
Trial history contributes to behavioral variability.

(A) Reward biases decision on subsequent trials. Plot indicates psychometric performance on trials following correct right (red) and correct left (blue) trials. Black line indicates the mean. Data …

https://doi.org/10.7554/eLife.11308.023
Head-restrained and unrestrained rats exhibit comparable task performance.

(A) Schematic of the accumulation of evidence task during head restraint. A rat initiates a behavioral trial by inserting his headplate into a custom headport along one wall of an operant training …

https://doi.org/10.7554/eLife.11308.024

Download links