Contrast thresholds reveal different visual masking functions in humans and praying mantises

Abstract

Recently, we showed a novel property of the Hassenstein-Reichardt detector, namely that insect motion detection can be masked by 'undetectable' noise, i.e. visual noise presented at spatial frequencies at which coherently moving gratings do not elicit a response (Tarawneh et al., 2017). That study compared the responses of human and insect motion detectors using different ways of quantifying masking (contrast threshold in humans and masking tuning function in insects). In addition, some adjustments in experimental procedure, such as presenting the stimulus at a short viewing distance, were necessary to elicit a response in insects. These differences offer alternative explanations for the observed difference between human and insect responses to visual motion noise. Here, we report the results of new masking experiments in which we test whether differences in experimental paradigm and stimulus presentation between humans and insects can account for the undetectable noise effect reported earlier. We obtained contrast thresholds at two signal and two noise frequencies in both humans and praying mantises (Sphodromantis lineola), and compared contrast threshold differences when noise has the same versus different spatial frequency as the signal. Furthermore, we investigated whether differences in viewing geometry had any qualitative impact on the results. Consistent with our earlier finding, differences in contrast threshold show that visual noise masks much more effectively when presented at signal spatial frequency in humans (compared to a lower or higher spatial frequency), while in insects, noise is roughly equivalently effective when presented at either the signal spatial frequency or lower (compared to a higher spatial frequency). The characteristic difference between human and insect responses was unaffected by correcting for the stimulus distortion caused by short viewing distances in insects. These findings constitute stronger evidence that the undetectable noise effect reported earlier is a genuine difference between human and insect motion processing, and not an artefact caused by differences in experimental paradigms.

Keywords: Masking; Motion detection; Praying mantis; Reichardt detector; Visual noise.

Fig. 1.

The Elementary Motion Detector (EMD). The spatial input from two identical Gaussian filters separated by Δx is passed through high and low pass temporal filters (HP and LP, respectively). The LP output in each subunit is cross-correlated with the HP output from the other subunit using a multiplication stage (M) and the two products are then subtracted to produce a direction-sensitive measure of motion.

Fig. 2.

Masked grating stimulus conditions used in Experiment H1. Each column represents one stimulus condition. Top row shows still frames of each condition, while middle and bottom rows show corresponding space-time plots and Fourier spatio-temporal amplitude spectra, respectively. In these plots, the signal contrast was set to 0.2. The cartoons at the top represent the conditions graphically and are used in subsequent figures for easy reference (signal is the upwards pointing arrow and noise is the coloured rectangle). The conditions are also labelled using the format Signal+Noise (S+N), where each of S and N is either H or L, indicating high and low spatial frequencies, respectively. For example, L+H indicates low frequency signal and high frequency noise.

Fig. 3.

Human motion detection contrast thresholds for different combinations of signal and noise frequencies (measured in Experiment H1). Bars show mean contrast detection thresholds (n=4) and error bars show ±s.e. of the mean. Horizontal brackets indicate threshold pairs that differ significantly (paired t-test, *P≤0.05 and **P≤0.01). Results show that each of the two signals frequencies 0.4 (blue) and 2 cpd (green) was masked significantly higher by same-frequency noise compared to different-frequency noise. Stimuli icons (below bars) and labels (above bars) use the notation introduced in Fig. 2.

Fig. 4.

Responses, fitted psychometric curves and detection thresholds of a single mantis (measured in Experiment M1). Circles show optomotor response rates (i.e. proportion of trials on which the mantis was coded as moving in the same direction as the signal grating) as a function of signal grating contrast. Error-bars are 95% confidence intervals calculated from simple binomial statistics. Red curves show fitted psychometric function (Eqn 4); red vertical lines mark contrast threshold. (A,C,E) Low-frequency signal (i.e. 0.04 cpd); (B,D,F) high-frequency signal (i.e. 0.2 cpd). Insets at the bottom right corner of each panel indicate signal and noise frequencies as in Fig. 2. (A,B) No noise: stimulus is a pure drifting luminance grating. (C,D) Low-frequency noise, i.e. added to the drifting signal grating is a grating of 0.04 cpd for which phase is updated randomly on every frame. (E,F) High-frequency noise. The data plotted in this figure are all from a single individual (mantis F11) and were measured in Experiment M1.

Fig. 5.

Mantis motion detection contrast thresholds for different combinations of signal and noise frequencies (measured in Experiment M1). Bars show mean contrast detection thresholds (n=6) and error bars show ±s.e. of the mean. Horizontal brackets indicate threshold pairs that differ significantly (paired t-test, *P≤0.05 and **P≤0.01). Stimuli icons (below bars) and labels (above bars) use the notation introduced in Fig. 2. Results show that the 0.2 cpd signal was masked to similar degrees by noise at either frequency, while the 0.04 cpd signal was masked more strongly by the 0.04 cpd noise.

Fig. 6.

Mantis motion detection contrast thresholds for different combinations of signal and noise frequencies (measured in Experiment M2). Bars show mean contrast detection thresholds (n=6) and error bars show ±s.e. of the mean. Horizontal brackets indicate threshold pairs that differ significantly (paired-sample t-test, *P≤0.05 and **P≤0.01). Stimuli icons (below bars) and labels (above bars) use the notation introduced in Fig. 2. The results show the same qualitative differences observed in Experiment M1 (Fig. 5): the 0.2 cpd signal is masked to similar degrees by noise at either frequency, while the 0.04 cpd signal is masked more strongly by 0.04 cpd noise. This similarity excludes the possibility that mantis and human results were different because stimuli appeared spatially distorted to mantises.

Fig. 7.

The spatial contrast sensitivity function of mantis optomotor response. The spatial frequencies used for signal and noise in experiments M1/M2 (0.04 and 0.2 cpd) are indicated on the plot using green/blue vertical lines. Contrast sensitivity data points are from Nityananda et al. (2015) and were corrected to adopt the same notation for converting between pixels and visual degrees as in Experiment M1 [i.e. averaging over screen width, instead of a single spatial period as in Nityananda et al. (2015)].

Fig. 8.

Masked grating stimulus conditions used in Experiment M2. Each column represents one stimulus condition. Top row shows still frames of each condition, while middle and bottom rows show corresponding space-time plots and Fourier spatio-temporal amplitude spectra, respectively. In these plots the signal contrast was set to 0.1. These stimuli conditions are similar to their correspondents in Experiment H1 (Fig. 2) but were modified in three ways: (1) they were limited to the central 85 degrees of the visual field, (2) they were corrected for spatial distortion by introducing a non-linear horizontal transformation, and (3) their noise was restricted to a single spatial frequency. Stimuli icons and labels (top row) use the notation introduced in Fig. 2.