Color diversity judgments in peripheral vision: Evidence against "cost-free" representations

Abstract

Is visual perception "rich" or "sparse?" One finding supporting the "rich" hypothesis shows that a specific visual summary representation, color diversity, is represented "cost-free" outside focally-attended regions in dual-task paradigms [1]. Here, we investigated whether this "cost-free" phenomenon for color diversity perception extends to peripheral vision. After replicating previous findings and verifying that color diversity is represented "cost-free" in central vision, we performed two experiments: in our first experiment, we extended the paradigm to peripheral vision and found that in minimally-attended regions of space, color diversity perception was impaired. In a second and final experiment, we added confidence judgments to our task, and found that participants maintained high levels of metacognitive awareness of impaired performance in minimally-attended visual areas in the periphery. These findings provide evidence that color perception may be partially attention-dependent in peripheral vision, and challenge previous views on both sides of the rich vs. sparse debate.

Fig 1. Experiment 1: Testing “cost-free” color perception in peripheral vision.

On each trial, following a brief fixation cross, participants received a 200ms cue (a curved white arc) which alerted them to the region of space that they needed to attend to. Following the cue, 24 colored letters appeared in a circular pattern for 300ms in the visual periphery. As shown by the inset, these letters were presented approximately 21 degrees away from fixation on our large projection screen. Following a blank interval of 900ms, participants then answered three questions: The cued letter recall, a judgment of whether the color diversity (in the cued or uncued regions) was either low or high, and their subjective color perception for the same region as the color diversity judgment. On a small proportion of trials, the white fixation cue was accompanied by a letter placed at fixation, and subjects were immediately asked to report the letter identity. These “catch” trials were randomly interleaved throughout the experiment to ensure that subjects were incentivized to always fixate on the cross.

Fig 2. Experiment 1: Average accuracy across subjects for letter recall and color diversity judgments.

The mean of all subject averages is shown, with error bars showing SEM across subjects. Dotted lines represent chance-level performance. (A) Letter Recall Task. The average accuracy across subjects for the letter recall task was significantly higher when the color diversity question focused on the cued region of the circle, compared to when this question queried the uncued region of the circle. (B) Color diversity judgments. The average accuracy for color diversity was higher for cued regions of the circle compared to uncued regions of the circle.

Fig 3. Subjective ratings for color diversity judgments in Experiment 1.

(A) Average subjective rating for color diversity judgments for the cued and uncued rows. Subjective ratings for uncued rows were significantly lower than cued rows. (B) Proportion of trials for each subjective rating type. The main difference between ratings for the cued and uncued trials were differences in the frequencies of the “1” and “3” responses. Error bars show SEM across subjects.

Fig 4. Measures of metacognition for color diversity judgments in Experiment 2.

The mean of all subject averages is shown, with error bars showing SEM across subjects. (A) d’ for color diversity judgments. (B) Meta-d’ for color diversity judgments. This measure was significantly higher for cued compared to uncued regions of the circle. (C) The M-ratio. This measure of metacognitive efficiency was nearly identical for cued and uncued regions of the circle. Note that the high p-value is from Wilcoxon test.

Fig 5. Confidence for correct and incorrect color diversity judgments in Experiment 2.

(A) Average confidence for cued and uncued color diversity judgments, separated by whether the judgment is correct or incorrect. As can be seen in the figure, in general, confidence was higher for correct than incorrect judgments, and this was true for both cued and uncued regions of the circle. (B). Average accuracy across color diversity judgments with different confidence levels, for cued and uncued regions. In general, accuracy increased as confidence increased for color diversity judgments in both cued and uncued regions. Error bars show SEM across subjects.