The functional neuroanatomy of episodic retrieval: using neuroimaging to understand the computational processes underlying human memory

Abstract

The medial temporal lobe (MTL) is essential to episodic memory through its role in the reactivation of past experience. Cortical networks provide top-down influence on the MTL, influencing the manner in which information can be retrieved. While neuroimaging investigations of human memory have characterized the functional correlates of episodic retrieval, the coordination of MTL systems during self-guided memory search is poorly understood. Using functional neuroimaging to estimate neural activation during variants of the free-recall paradigm, this dissertation demonstrates that the posterior MTL is critically involved in internally-directed memory search, through its interactions with distributed cortical systems. First, I characterize the engagement of a cortico-hippocampal network during memory search. I then demonstrate increased functional connectivity between this network and multiple frontoparietal systems, identifying neural mechanisms that may potentially reflect top-down control of memory search. Next, I relate activation of the posterior MTL to the process of episodic recollection through comparison of activity during free recall and source recognition tasks. I additionally demonstrate common activation in dorsal frontoparietal networks during free recall and processing of item familiarity. To characterize the computations mediated by the MTL, I develop a neuro-cognitive model of free recall. Retrieved-context theories propose that temporal context, a slowly integrating representation of the recent past, cues the hippocampus during retrieval. I link activation of the MTL to the process of temporal reinstatement, predicting the temporal organization of recall. Next, I use a variant of the free-recall paradigm in which memory is disrupted prior to retrieval, to test the capacity of large-scale cortical networks to control episodic memory. I show that a frontoparietal control network (FPCN) functionally couples with the MTL when memory is disrupted. Using a neurally informed computational model of recall, I demonstrate that activation of the FPCN and posterior MTL predict when individuals will overcome distraction by reinstating contextual information to guide memory search. Taken together, these findings demonstrate how the MTL supports memory search through the reinstatement of contextual information, a process that is coordinated through top-down signals from frontoparietal networks.