Figure 1
Experimental design.
Behavioural data were collected during a simple visual detection task. Each trial started with a fixation cross, followed by the appearance of a probabilistic cue at the centre of the screen. Subsequently, a checkerboard, either containing or not grey circles at the titrated contrast level, appeared at the bottom left corner of the monitor for a duration of 0.06 seconds. Following stimulus presentation, two screens were displayed, the first asking for type 1 (target detection) and the second for type 2 (confidence rating) responses respectively. The presented cues took the form of a blue bar, filled with different levels of red ink in the proportion of 33, 50 and 67% of the bar. The degree of red filling of the blue bar was related with the probability of target presence, with 33% indicating low probability and 67% high probability of target, corresponding to conservative and liberal conditions, respectively while the half-filled (50%) bar representing the random target probability occurrence, being thus uninformative of target presence vs. absence.
Figure 2
Interaction between prior information and response correctness in shaping confidence levels.
Confidence levels were influenced by the accuracy of the responses, showing greater levels following correct relative to incorrect responses. Moreover, no difference between probability trials were observed for incorrect responses, while higher confidence was observed on low-probability trials compared to random and high probability trials for correct choices.
Figure 3
Congruency effect between prior information and stimulus presented.
Prior information influenced confidence levels based on the congruence between target probability and subsequent target presence. After catch trials, higher confidence was observed when the low-probability cue (33%) was presented compared to the high-probability one, whereas in target trials, the reverse trend was evident, with higher confidence linked to the high-probability cue (67%). Following the presentation of the random-probability cue (50%), no distinction in confidence levels between target and catch trials was observed.
Figure 4
Congruency effect between prior information and response provided.
Prior information impacted confidence levels depending on the congruence between target probability and the subsequent response given by the participant. When participants indicated target absence, greater confidence was linked with the low-probability condition (33%) rather than high-probability one (67%). Conversely, when participants reported target presence, higher confidence was associated with the high-probability condition. There was no discernible difference in confidence levels following responses given in the random-probability condition (50%).
Figure 5
Association between prior-dependent modulation of criterion and confidence.
It is observable a significant positive correlation between the prior-dependent adjustment of criterion and confidence, observed in both target-absent (A) and target-present (B) responses. Similarly, a positive association driven by prior information was also detected between criterion and confidence adjustment in target-present trials (C), but not in target-absent ones (D).