Imagery adds stimulus-specific sensory evidence to perceptual detection

Abstract

Internally generated imagery and externally triggered perception rely on overlapping sensory processes. This overlap poses a challenge for perceptual reality monitoring: determining whether sensory signals reflect reality or imagination. In this study, we used psychophysics to investigate how imagery and perception interact to determine visual experience. Participants were instructed to detect oriented gratings that gradually appeared in noise while simultaneously either imagining the same grating, a grating perpendicular to the to-be-detected grating, or nothing. We found that, compared to both incongruent imagery and no imagery, congruent imagery caused a leftward shift of the psychometric function relating stimulus contrast to perceptual threshold. We discuss how this effect can best be explained by a model in which imagery adds sensory signal to the perceptual input, thereby increasing the visibility of perceived stimuli. These results suggest that, in contrast to changes in sensory signals caused by self-generated movement, the brain does not discount the influence of self-generated sensory signals on perception.

Figure 1.

Experimental design and hypotheses. (A) Experimental paradigm. At the start of each block, participants were instructed whether they would be detecting left- or right-tilted gratings and what they had to imagine this block: nothing, a left-tilted grating, or a right-tilted grating. Trials consisted of 200-ms fixation followed by 2 s of dynamically fluctuating noise in which a stimulus gradually ramped up until a certain visibility level. The task was to indicate whether a grating was present or not. After each block, participants were asked what they had imagined during this block to check whether they accurately followed instructions. (B) Hypotheses. Imagery might increase or decrease perceptual presence response via different mechanisms. Left: a response bias would be reflected in a change in guess rate and mean. Middle: a change in sensory sensitivity would be reflected in a change in slope. Right: a subtraction or addition to the sensory signal would be reflected in only a change in mean.

Figure 2.

Results. (A) Psychometric function per condition. Solid points represent proportion presence trials at the different visibility levels for each condition averaged over participants. Vertical lines reflect SEM for these points. (B) Curve parameter estimates. Top: mean, the signal value that is associated with a 0.5 probability of reporting stimulus presence. Middle: variance/slope, the sensitivity of presence responses to changes in signal. Bottom: guess rate, the proportion of presence responses for zero signal. *p < 0.05; **p < 0.01; ***p < 0.001.