Internal noise measures in coarse and fine motion direction discrimination tasks and the correlation with autism traits

Abstract

Motion perception is essential for visual guidance of behavior and is known to be limited by both internal additive noise (i.e., a constant level of random fluctuations in neural activity independent of the stimulus) and motion pooling (global integration of local motion signals across space). People with autism spectrum disorder (ASD) display abnormalities in motion processing, which have been linked to both elevated noise and abnormal pooling. However, to date, the impact of a third limit-induced internal noise (internal noise that scales up with increases in external stimulus noise)-has not been investigated in motion perception of any group. Here, we describe an extension on the double-pass paradigm to quantify additive noise and induced noise in a motion paradigm. We also introduce a new way to experimentally estimate motion pooling. We measured the impact of induced noise on direction discrimination, which we ascribe to fluctuations in decision-related variables. Our results are suggestive of higher internal noise in individuals with high ASD traits only on coarse but not fine motion direction discrimination tasks. However, we report no significant correlations between autism traits and additive noise, induced noise, or motion pooling in either task. We conclude that, under some conditions, the internal noise may be higher in individuals with pronounced ASD traits and that the assessment of induced internal noise is a useful way of exploring decision-related limits on motion perception, irrespective of ASD traits.

Figure 1.

(a) Modeled threshold versus external noise curves, as measured with equivalent noise paradigms. As external (stimulus) noise is increased, performance thresholds (i.e., observed noise) rise. When compared to a baseline (with a certain level of additive but no induced noise), the following effects are observed: Reducing motion pooling leads to a uniform upward shift in the curve; increasing additive noise increases thresholds but only at low levels of external noise; and induced noise increases thresholds, especially at high levels of external noise. The vertical dashed line identifies the elbow in the baseline curve and quantifies the level of additive noise according to the equivalent noise paradigm. (b) Internal noise versus external noise curves, as obtained with a double-pass paradigm. When expressed in terms of the total amount of internal noise, there are clear differences between the different manipulations. Increasing additive noise elevates the internal noise by the same amount independent of the external noise. When induced noise is included, internal noise shows a strong dependence on external noise. Induced noise is the only variable that causes an increase in internal noise in the internal versus external noise plot. In the original approach by Burgess and Colborne (1988), motion pooling was not taken into account. When using this approach, motion pooling scales the curve up or down equally at all external noise levels (causing misestimated noise levels, as shown in this figure). However, when explicitly including motion pooling in the model, motion pooling would have no effect on internal noise estimates (as per our method).

Figure 2.

Screenshot of visual motion task stimuli (a) and response screen (b). Participants viewed the stimuli and then gave a response by clicking on a wedge in the response screen indicating whether the dots were moving clockwise or anticlockwise of vertical and how confident they were in their response.

Figure 3.

Relationship between accuracy and consistency, dependent on different ratios of  σ_int/σ_ext. (a) Colored lines were calculated using a wrapped Gaussian distribution, and the dashed gray lines are based on Gaussian distributions used by Burgess and Colborne (1988). The line for σ_int/σ_ext = ∞ is the case without external noise. (b) Schematic of how data were analyzed for two example data points. Internal noise for the data falling above the  σ_int/σ_ext = ∞ curve (e.g., square) was estimated by finding the nearest point on the σ_int/σ_ext = ∞ curve, and the internal noise associated with that point was taken as the measured internal noise. For all other data (circle), the best fitting line determined σ_int/σ_ext.

Figure 4.

Observed external noise and motion pooling. This plot has the same layout as Figure 3 but also shows iso-external noise lines: the accuracy and consistency expected for a given external noise level without motion pooling (displayed on the right). The black dots are experimental data for one participant, and the external noise for each condition is indicated within the dot. The external noise for each condition is higher than the iso-external noise line on which it falls, suggesting that the observed/effective external noise was lower than the actual external noise, an indication that motion pooling was occurring.

Figure 5.

Mean behavioral performance. (a) Accuracy, (b) consistency, and (c) confidence decreased as external noise level increased but remained significantly above chance level at all noise levels. Error bars are standard errors of the mean calculated over participants.

Figure 6.

Dependence of estimated internal noise on external noise for the coarse discrimination task. (a) Average results over all participants. (b) Histogram depicting the distribution of AQ scores. (c) Internal noise after a median split for low- and high-AQ groups. Error bars indicate ±SEM.

Figure 7.

Correlations for the coarse discrimination task. (a) Additive internal noise, (b) induced noise factor m, and (c) motion pooling, dependent on AQ. One outlier was removed in (c).

Figure 8.

Mean behavioral performance in the fine discrimination task. (a) Accuracy, (b) consistency, and (c) confidence all decreased as external noise increased but remained significantly above chance at all noise levels. Error bars show the standard errors of the mean over participants.

Figure 9.

(a) Dependence of mean (±SEM) internal noise on external noise for the fine discrimination task, averaged over participants. (b) Histogram depicting the distribution of AQ scores. (c) Internal noise after a median split for low- and high-AQ groups. Data points that were estimated as 0 internal noise were ignored.

Figure 10.

Estimates of (a) additive internal noise, (b) the induced noise factor m, and (c) motion pooling from the fine direction estimation task plotted against AQ.

Figure 11.

Simulation results for different fitting procedures for the fine direction discrimination experiment. Results show histograms of estimated model parameters from 500 simulated runs. The red lines indicate the true values. (a) Full fit in which additive noise, induced noise, and motion pooling were all individually fit. (b) Fit without pooling (WOP) in which additive noise and induced noise were fitted, but the pooling parameter was estimated directly from the data. (c) Estimated induced noise, the median induced noise value for the five largest external noise conditions. (d) Estimated induced noise taken from a single fit through the five largest external noise conditions.

Figure 12.

Estimated motion pooling sample size. (a) Estimates based on Equation 2 from one of the simulations. Estimates for the fine direction discrimination task were accurate for larger external noise values used in the experiment. (b) Boxplots for the coarse direction discrimination task; these estimates were accurate for the lower external noise values used in the experiment. (c, d) Experimentally obtained motion pooling estimates for fine (c) and coarse (d) discrimination tasks. The box of the boxplots contains the 25th to 75th percentiles; the whiskers show the most extreme values not considered outliers, and the red markers show the outliers.