Traces in time and space: How the past guides attention

To navigate effectively in a dynamic visual world, it is crucial to manage constantly changing sensory inputs and coordinate timely responses. In everyday life, humans achieve this efficiency by learning to predict where and when visual information will appear and by preparing actions in advance. This thesis explores how visual statistical learning of time-event regularities supports adaptive visual attention and motor processes, enhancing performance in distraction-rich environments. Chapter 2 investigates whether visual attention can be guided by learned time-target location regularities. Participants performed a visual search task with such regularities. Following a short interval, the target was more likely to appear in one location, while after a long interval, it was more likely in the opposite location. Results showed that, even without awareness, performance was better when targets appeared at the probable location after its associated interval. This indicates that attention was directed more effectively to expected target locations based on timing. The chapter demonstrates that time-target location regularities can be learned implicitly, allowing the attentional priority map to be dynamically adjusted, improving target selection. Chapter 3 examines whether such learning depends on target saliency. Participants completed a visual search task with time-target location regularities, while target saliency was manipulated (salient vs. nonsalient). Results indicated that search efficiency improved at temporally valid locations, independent of saliency. This suggests that participants learned to anticipate target locations based on timing regardless of distinctiveness. Thus, target saliency is not necessary for learning time-target associations, underscoring the robustness of statistical learning in guiding attention independently of stimulus-driven capture. Chapter 4 explores how the system handles irrelevant distractions to maintain efficiency. It tests whether visual attention can be tuned by learned time-distractor location regularities to mitigate interference. In a visual search task, a distractor occasionally appeared alongside the target. Unbeknownst to participants, distractors followed spatiotemporal patterns, occurring more frequently in one location after a short interval and another after a long interval. Results showed interference was reduced when distractors appeared at probable locations and intervals compared to nonassociated ones. This suggests participants learned time-distractor regularities and used them to direct attention away from likely distractor locations at expected times. The chapter demonstrates that statistical learning can also reduce distractions, dynamically adjusting attentional maps to filter irrelevant stimuli. Chapter 5 extends these findings by exploring time-event learning beyond simple binary contexts, examining how complex regularities influence both attention and action. Participants completed a speeded choice reaction task involving time-target location regularities, time-motor response regularities, or both. Events were linked to distinct timing patterns, with one following an exponential and the other an anti-exponential distribution across four intervals. Results showed enhanced performance when events aligned with their patterns, with improvements persisting in a transfer phase after regularities were removed and participants were informed. These findings indicate that complex time-event associations can be robustly learned and retained across multiple processing levels, enhancing efficiency in both perception and motor response. Chapter 5 underscores that statistical learning supports attentional guidance and motor preparation through a long-term, implicit associative process across perceptual and motor domains. Together, this thesis illustrates the impact of visual statistical learning on shaping attention and motor readiness. It shows how cognitive systems leverage past experiences with time-event regularities to anticipate and respond efficiently to future events in complex, dynamic environments.