Abstract
During visual search we quickly learn to attend to an object’s probable location, efficiently sifting through clutter from the visual world to find our target. Research has supported that this process is facilitated when target-location learning is based on hippocampal-dependent spatial contextual associations (CC, contextual cueing) or striatal-based probabilistic regularities (LPL, location probability learning). Here, we tested how these different types of learning aid the utilization of established memories. In two online experiments, participants searched for targets within scenes. Depending on the scene category, the target consistently appeared at a specific location (CC), within a hemifield (LPL), or was unpredictable (random). In Experiment 1, 54 participants were subsequently tested on their memory for the hemifield and the specific location of the learned targets. Participants showed enhanced recall accuracy for target hemifield and specific target location in both LPL and CC conditions. However, when learning performance was low (low accuracy/high reaction time), predominantly LPL facilitated memory for target hemifields, and when learning performance was high, CC facilitated memory for specific target locations. In Experiment 2, after learning, 54 participants were tested on their ability to orient attention to targets flashed either in a learned specific location or hemifield. We found greater orienting benefits for CC compared to LPL, as measured by reaction time. Together, we demonstrate that contextual and probabilistic learning systems provide utility for future retrieval of learned associations, but how these systems promote memory retrieval may be related to the quality of encoding. Further, after comparable learning conditions, attentional orienting seems more profoundly guided by contextual, compared to probabilistic regularities. Our work suggests that a more nuanced view of how these memory systems cooperate and/or compete to guide adaptive behavior is necessary.