The concept of target templates is a core construct in all models of attention, guiding attention during search and acting as a “template” for making target match decisions. Previous work has generally focused on how the identity of a known target is stored in the target template and the consequences of failures to maintain that representation in memory. Few, if any, have tested how the target template adapts to uncertainty in target features and makes predictions about how the target will look in the future. In this doctoral thesis, I present a series of four experiments that address how predictive uncertainty shapes representations of target templates, on both cognitive and neural levels. I show that observers readily learn statistical knowledge about targets and use this knowledge to make predictions about the likelihood of various target features. These predictions are then used to set priority for features which minimize uncertainty about the upcoming target, biasing eye-movements and reducing decision times when participants are certain about how the target will look. At the neural level, I show that feature uncertainty is coded in dorsolateral prefrontal cortex and inferior frontal junction, supporting search behavior for uncertain targets. These findings contribute to our understanding of how attentional templates adapt to statistical knowledge about uncertain target features and provide insights into theories of attention and perceptual learning.

Significance statement

Multisensory experiences are ubiquitous in our everyday lives and impact what sensory information we notice, pay attention to, and remember. However, many areas of cognitive psychology focus on the senses individually, and/or use simplistic versions of real objects. This makes it difficult to understand whether, and how, laboratory findings can explain cognitive processing in the real world. This dissertation investigates how naturalistic object properties, including multiple sensory modalities, semantic information, and dynamic motion, contribute to sensory processing and memory formation. Chapter 2 examines the impact of task-irrelevant, semantically congruent sounds on visual recognition memory. Through a series of experiments, it demonstrates that congruent object-sound pairings facilitate recollection-based recognition and promote the formation of multisensory memories. These findings underscore the importance of considering multisensory interactions in developing models of memory applicable to real-world settings. Building on these insights, Chapter 3 investigates how multisensory object processing affects memory for nearby visual objects and scene contexts. While the presence of audiovisual objects at encoding did not significantly benefit memory for nearby visual objects, it did improve recall of the environmental context. These results highlight the broader influence of multisensory processing on episodic memory formation beyond individual objects. Chapter 4 explores the audiovisual ventriloquist effect using realistic stimuli in virtual reality. In this study, we found that animated, semantically congruent audiovisual stimuli show enhanced spatial ventriloquism at small disparities relative to the simplistic stimuli frequently used in laboratory studies of multisensory integration. The study emphasizes the role of stimulus realism and dynamic motion in audiovisual integration. Collectively, this research advances our understanding of how multisensory experiences shape memory and perception in naturalistic settings.

Attention as an umbrella term covers a vast array of topics. In Chapter 1, we review the relevant background on selective attention, as it is a vital function for humans to navigate the stimulus-rich environments they are immersed in and how it relates to object recognition. In Chapter 2, we explore the role of target templates that are powerful heuristics for integrating remembered concepts with novel percepts in a task-efficient manner. Target-distractor similarity and linear separability have previously been shown to moderate properties of the target template representation in memory for unidimensional objects; we sought to extend these findings in multidimensional objects (each dimension of which was independently manipulated) by recruiting participants to perform simultaneous visual search and memory probe tasks. Results showed that target-distractor similarity moderated search efficiency and was associated with an off-veridical memory bias and an altered method of extracting information. And in Chapter 3, we review established and experimental interventions for children and adolescents with attention deficit/ hyperactivity disorder to improve this population’s wide array of atypical and adverse symptoms, including selective attention deficits.