Memory serves as a critical bridge between our present and our past, allowing previously formed associations to guide our present decision-making and inform our expectations about what is to come. Our ability to learn and retrieve certain types of associative knowledge, as well as to make projections about the world based on this knowledge, depends on the neural circuitry of the hippocampus, along with surrounding structures that comprise the medial temporal lobe (MTL) (Squire, 2009b). Declarative memory has been posited to rely on processes such as pattern completion and separation (Yassa & Stark, 2011), sequence encoding (MacDonald, Lepage, Eden, & Eichenbaum, 2011), and mismatch detection (Kumaran & Maguire, 2006). These processes enable memory to be flexibly encoded and retrieved so that it can inform all aspects of experience. This dissertation address three specific questions regarding how subfield functions work within a broader system of goals and motivations to facilitate flexible declarative memory. The project described in Chapter 3 focuses on the different degree to which discrimination and generalization responses, previously observed in hippocampal subfields in response to close lure stimuli, is affected by task demands. Using high-resolution functional magnetic resonance imaging (hr-fMRI), we find diverging sensitivity to changes in stimuli and task demands in the responses within the left CA1 and CA3DG subfields. CA3DG discriminates old visual stimuli from new stimuli, irrespective of the task-relevant context. In contrast, CA1 responds to incorrectly paired repeats as it would to novels, while generalizes to different exemplars of the expected image. In Chapter 4, we use hr-fMRI investigate the impact of value on memory and show that items assigned a low reward value exhibit reduced encoding-retrieval neural similarity than either items assigned high reward value, or the similarity between items at encoding and unvalued items at retrieval. These results suggest that low value items may be processed in a less consistent manner within MTL regions than higher value items during encoding, possibly reflecting different encoding processes due to assigned values. Chapter 5 describes a set of behavioral studies aimed at evaluating whether mnemonic mismatches can lead to improved subsequent recollection of the item that violated memory based predictions. We show that when an item triggers a mismatch that item is recalled at greater rates than items that did not constitute a mismatch. Our results also show that this putative mismatch-induced encoding benefit is susceptible to disruption by numerous goal-related or motivational factors.
展开▼
