Fig 1.
Eye matching task stimuli and conditions.
a. Morphing procedure and stimulus creation. Two morphed eye regions are selected from the morphing continuum and pasted on two distinct face contexts. b. Same, Different and Isolated context conditions. Bracket shape (round, square) defines distinct face contexts, percentage values define morphing percentage in the eye region. Target regions within a pair had 0%, 24%, 36%, 53% or 80% dissimilarity. It can be seen here that when the target regions are the same (0% dissimilarity) and context differs, the contextual modulations are stronger compared to when target regions differ largely (80% dissimilarity).
Fig 2.
Same, different, and isolated context conditions of the contrast detection task.
Participants were asked to locate the horizontal grating appearing in one of the four target regions. It can be seen here that although target contrast is identical in all three examples, target appears to be less visible in Same condition compared to other two conditions.
Fig 3.
Eye matching task trial sequence.
An example of 0% eye region dissimilarity and Different context condition trial. The participant sees a fixation cross for 500 ms at the beginning of each trial, followed by a blank screen. Afterwards, participant sees the first stimulus for 500 ms, then the noise mask for 200 ms, and finally, the second stimulus. The second stimulus stays on the screen until the participant indicates whether the eye regions were “Same” or “Different” between the two stimuli by button press, followed by a blank screen before the next trial starts.
Fig 4.
Data and psychometric functions of a representative participant.
The dependent variable in the eye matching task is the proportion of ‘Different’ responses, and accuracy in the contrast detection task. Dots represent actual data (varying in size depending on frequency of presentation) and curves represent the model fits, with 95% HDI shaded ribbons. Vertical lines mark the “threshold” for a given condition (at .5 in the eye matching task, at .75 for the contrast detection task).
Fig 5.
A. Z-scaled threshold estimates in eye matching (left) and contrast detection (right) tasks. Each dot represents the z-scaled threshold estimate for a given task and condition for a participant. The top, middle and bottom lines of boxes correspond to the first, second (median) and third quartiles (25th, 50th and 75th percentiles) respectively. The upper and lower whiskers extend until 1.5* Inter-Quartile Range. Faded grey lines connect the thresholds in respective conditions for a given participant. Bold black dots and lines display group means. B. Posterior odds of differences across the three tasks. The density plots display the posterior odds of the threshold differences. 95% interval of the distributions did not overlap with zero, therefore all differences were non-negligible.
Fig 6.
Correlation of contextual modulation magnitudes across tasks.
Contextual modulation magnitude values are the product of regression of regressions method for each task. Blue line indicates the regression line and shaded area covers the standard error bounds.
Fig 7.
Correlation of the individual threshold profiles across tasks.
Each dot represents a Fisher-Z transformed Pearson Correlation Coefficient value of an individual’s threshold values from the corresponding comparison. The top, middle and bottom lines of boxes correspond to the first, second (median) and third quartiles (25th, 50th and 75th percentiles) respectively. The upper and lower whiskers extend until 1.5* Inter-Quartile Range. Black diamonds display means.