Estibaliz San Anton

The reversal of perceptual and motor compatibility effects differs qualitatively between metacontrast and random-line masks

In masked priming tasks, participants typically respond faster to compatible than to incompatible primes, an effect that has been dubbed as the positive compatibility effect (PCE). However, when the interval between the prime and the mask is relatively long, responses are faster to incompatible than to compatible primes. This inversion is called the negative compatibility effect (NCE). Two main origins of the NCE have been proposed. The object-updating theory holds that when the masks share stimulus features with the primes, both perceptual and motor processes generate an NCE. As an example, for masks composed of overlaid left and right prime arrows, the NCE is thought to be positive priming induced by the arrow of the mask pointing in the opposite direction of the prime. In contrast, the motor inhibition theories hold that the origin of the NCE is purely motor and can be demonstrated when masks do not share features with primes. To test both hypotheses, the present study aims at delineating the respective contributions of perceptual and motor components of the NCE in the context of different types of masks. Consistent with the object-updating hypothesis, we found both perceptual and motor NCEs at the long SOA with metacontrast masks (with internal contours corresponding to left and right overlaid arrows). Consistent with the motor inhibition hypothesis, we found motor NCE but no perceptual NCE at the long SOA with random-line masks (containing no prime features). The study thus suggests that the origin of the NCE depends on the type of mask.

Unconscious associative learning with conscious cues

Abstract Despite extensive research, the very existence of unconscious learning in humans remains much debated. Skepticism arises chiefly from the difficulty in assessing the level of awareness of the complex associations learned in classical implicit learning paradigms. Here, we show that simple associations between colors and motion directions can be learned unconsciously. In each trial, participants had to report the motion direction of a patch of colored dots but unbeknownst to the participants, two out of the three possible colors were always associated with a given direction/response, while one was uninformative. We confirm the lack of awareness by using several tasks, fulfilling the most stringent criteria. In addition, we show the crucial role of trial-by-trial feedback, and that both the stimulus–response (motor) and stimulus–stimulus (perceptual) associations were learned. In conclusion, we demonstrate that simple associations between supraliminal stimulus features can be learned unconsciously, providing a novel framework to study unconscious learning.

Saving the Perruchet effect: A role for the strength of the association in associative learning

In a partial reinforcement schedule where a cue repeatedly predicts the occurrence of a target in consecutive trials, reaction times to the target tend to decrease in a monotonic fashion, while participants’ expectancies for the target decrease at the same time. This dissociation between reaction times and expectancies—the so-called Perruchet effect—challenges the propositional view of learning, which posits that human conditioned responses result from conscious inferences about the relationships between events. However, whether the reaction time pattern reflects the strength of a putative cue-target link, or only non-associative processes, such as motor priming, remains unclear. To address this issue, we implemented the Perruchet procedure in a two-choice reaction time task and compared reaction time patterns in an Experimental condition, in which a tone systematically preceded a visual target, and in a Control condition, in which the onset of the two stimuli were uncoupled. Participants’ expectancies regarding the target were recorded separately in an initial block. Reaction times decreased with the succession of identical trials in both conditions, reflecting the impact of motor priming. Importantly, reaction time slopes were steeper in the Experimental than in the Control condition, indicating an additional influence of the associative strength between the two stimuli. Interestingly, slopes were less steep for participants who showed the gambler’s fallacy in the initial block. In sum, our results suggest the mutual influences of motor priming, associative strength, and expectancies on performance. They are in line with a dual-process model of learning involving both a propositional reasoning process and an automatic link-formation mechanism.