The human mind is rarely, if ever, inert. We are constantly barraged with external stimuli, but also experience a rich inner life of recollections, planning, and pondering which occur spontaneously throughout our waking hours. In this dissertation, I have aimed to investigate the cognitive mechanisms underpinning the segmentation, differentiation, and representation of event memories, acknowledging the ongoing dynamics of internal and external attention. I have taken as a paradigm the dynamic and flexible interplay between internal and external attention as a critical factor in memory encoding and organization. In Chapter 2, I present findings from a series of experiments which demonstrate the ways in which fluctuations in on-going attention (i.e. mind-wandering) act upon the later recollection of temporal order. In a similar vein Chapter 3 demonstrates how hierarchical contexts which change at different rates instrumentally decrement or benefit later recollection of temporal order memory. Building upon the idea of hierarchical and dynamic representations, in Chapter 4 I present a theoretical account of how attention and mind wandering may play an instrumental role in the alteration of memory organization and recollection in individuals with depression. In this chapter, I critique the canonical theory of overgeneral memory in depression and instead underscore basic long-term memory and attention processes to explain how memories may become blunted in mood disorders. Finally, in Chapter 5, I test this hypothesis by investigating how memory similarity may be derived from semantic representations to demonstrate how personal memory organization is changed in individuals with depression compared to healthy controls. These chapters highlight the theoretical and methodological advancements that have been and should continue to be made in the course of this dissertation.

The question of what role consciousness plays in memory is as old as modern memory research. While most studies have used techniques that briefly present the stimuli to study unconscious learning, these studies tend to show small effect sizes and are not always successful. This makes it difficult to compare the mechanisms of conscious and unconscious learning. As a result, a debate has ensued about whether learning the relationship between two events is possible without awareness. One thing proponents of both camps with opposite views on this matter agree on, is the need for better methods. This dissertation aims to test whether multi-voxel neuroreinforcement is one such method. Understanding unconscious learning is however only half of the problem, as conscious associative learning also needs investigation. To address these gaps, In Chapter 2, I show that unconscious neuroreinforcement can affect the relative status of a cue by affecting variability in choice behavior. In Chapter 3, I explore a form of associative learning called higher-order conditioning to understand how conscious learning unfolds, and show how two mechanisms that differ in their computational complexity can be utilized to learn and guide behavior. Lastly, Chapter 4 investigated whether the more computationally complex form of learning can be supported unconsciously and found further evidence in support of variability on choice behavior as a result of neuroreinforcement. These chapters highlight a need for using a range of experimental techniques to elucidate the mechanisms of conscious and unconscious learning.

Advancements in neuroimaging and brain stimulation techniques have provided unique opportunities to further understand the neural mechanisms of episodic memory retrieval. The act of retrieving information about a past experience is known to depend on the coordinated engagement of a broad networks of regions, including frontal lobe regions such as the rostrolateral prefrontal cortex (RLPFC) and medial temporal lobe areas such as the hippocampus (Cabeza & St. Jacques, 2007; Maguire & Mummery, 1999; Reynolds, McDermott, & Braver, 2006). Although much of the field’s extant knowledge has been derived from studies assessing memories formed in laboratory-based settings, the incorporation of life-logging technology – such as wearable digital camera devices – can assist with the nonintrusive photographic capture of everyday life events, which can later be employed as mnemonic probes. The experiments in this dissertation aim to assess the neural mechanisms mediating real-world episodic retrieval by employing naturalistic stimuli to elicit memories for personal experiences.

This dissertation begins with a broad overview of the behavioral and neural findings derived from memory experiments incorporating wearable camera technology, followed by novel examinations of the neural correlates underlying real-world events through the use of functional magnetic resonance imaging (fMRI) and high-definition transcranial direct current stimulation (HD-tDCS). Chapter 2 featured an in-depth review of prior applications of wearable digital cameras to behavioral and neuroimaging assessments of autobiographical memory retrieval as well as how their contributions expand knowledge of such processes to naturalistic settings. Chapters 3-5 report the results of a series of fMRI investigations examining recall of events from the real world and how they may differ across mnemonic features related to the original experiential source of the event, the recognition of the event based on previously encountering photographs of those experiences, and the temporal order of the event details. Chapter 3 found that dissociable patterns of neural activation were evoked in brain networks previously implicated in either autobiographical or laboratory-based memory retrieval (McDermott, Szpunar, & Christ, 2009), such that the autobiographical memory network was preferentially sensitive to whether or not the depicted events had been personally experienced, while the laboratory-based network was preferentially sensitive to whether or not photographs of the depicted events had been previously encountered. These findings suggest that these networks contribute to different retrieval processes and showcase how memories for first-hand experiences have distinctive neural signatures from memories for second-hand event knowledge. Chapter 4 focused on the hippocampus, with an emphasis on the division of labor along the hippocampal long-axis. The findings revealed that the posterior hippocampus was disproportionately sensitive to the source of the photographs, whereas the anterior hippocampus reacted more strongly to whether the photographs themselves had been previously seen, as well as whether their temporal order was intact. Chapter 5 assessed hemispheric differences in RLPFC responsivity to violations of temporal order during retrieval. The left RLPFC exhibited greater activation for temporal order violations only when events were novel, while the right RLPFC demonstrated greater activation for temporal order violations only when events had been previously encountered as photographs. These results suggest that the RLPFC is capable of differentially determining whether events are consistent with either prior schemas or memories. To further examine the left RLPFC and evaluate its causal involvement in mnemonic processes, Chapter 6 applied HD-tDCS methodology to this region in order to determine its impact on event recognition and temporal order processing. The targeted application of anodal current to the left RLPFC produced an increased likelihood of false recognition and – relative to sham stimulation – led to a shift in response bias, which may indicate the RLPFC’s role in memory monitoring. Together, these findings from fMRI and HD-tDCS experiments help clarify the contributions and characteristics of the neural substrate supporting episodic memory retrieval, particularly with regards to how these processes may occur in the real world.

Cognitive psychologists and neuroscientists traditionally take advantage of tightly controlled experimental studies to make claims about how the mind and the brain work – we design tasks and conditions to isolate a single component and use the results to make conclusions. However, it is becoming increasingly clear that as we move to understand more complex, naturalistic cognition, we need to consider the integration of cognitive processes. This dissertation probes how the integration of processes involved in human memory can help us better understand behavior. The first chapter begins by providing an overview of three domains of memory that are subserved by multiple components – declarative memory, memory for faces, and working memory. Chapter 2 investigates the bidirectional interactions of episodic and semantic memory by employing a paired associate learning task with semantically related and unrelated word pairs that additionally manipulates whether the word pairs are learned through either active retrieval practice or passive restudying. This paradigm, along with a novel extension of a multi-arrangement task (Kriegeskorte & Mur, 2012) to index semantic space before and after learning, shows that prior knowledge asymmetrically reshapes semantic space to make cue words more predictive of target words and testing reduces potential interference by repelling moderately related lure pairs away from the to-be-learned pair in semantic space. Chapter 3 shifts to focus on face memory. In this chapter, we use a large battery of behavioral tasks to identify four latent cognitive processes – face perception, episodic long-term memory, general intelligence and working memory. We show that these factors linearly combine to predict performance on two measures of face memory, one of which we designed to be a more ecologically valid index of personal identity memory. We also identify two cognitive profiles that differentially use working memory and face perception to accomplish our face memory task, highlighting how individual differences in cognitive ability may impact how processes interact to support behavior. In Chapter 4, we sought to explain individual differences in working memory performance, working memory capacity (WMC), and psychiatric outcomes using a wide range of structural and functional MRI measures. Our results highlight how working memory performance and WMC may be predicted by different aspects of functional and structural MRI that reflect distinct underlying processes. Taken together, the findings presented in this dissertation underscore the complexity of cognition. Expanding how we study the mind and brain to take multiple processes of cognition into account will allow us to better characterize and understand complex, real-life behavior.

The identification of neurocognitive correlates supporting enhanced memory has far reaching implications, from improving education and productivity to mitigating ailments caused by memory disorders. The work presented within this dissertation leverages behavioral paradigms, Virtual reality (VR), and neuroimaging tools to unveil the neural processes responsible for enhanced memory. The methodological tools used throughout this dissertation were drafted in an attempt to create an ecologically valid medium with theoretical extensions into education and rehabilitation. Of particular focus is how the parallel recruitment of reward and/or spatial processing systems during the encoding of information can serve to upregulate mnemonic processing and enhance memory. Specifically, Chapter 1 examines how memory-retrieval can be modulated by reward and whether individual differences in anatomical connectivity within reward processing and elaborative semantic encoding circuits, as measured with diffusion tensor imaging (DTI), are associated with value-induced modulation of memory.

The remaining chapters focus on the use of VR to implement the Method of Loci (MoL), the world’s most ancient and effective mnemonic. Behavioral and neuroimaging analyses were designed to test a hypothesized mechanism of action behind the MoL’s efficacy: explicit binding of information to the spatial scaffolding of an environment recruits neural systems supporting the encoding of space, which bolsters recall breadth and strength. To test this hypothesis, the dissertation first reviews the literature and presents VR as a way to increase the ecological validity of fMRI memory research (Chapter 2). Chapter 3 then investigates how the MoL can be implemented within VR and reveals which facets of the technique are most responsible for its potent impact on human memory. Chapter 4 extends the work in Chapter 3 by evaluating virtual strategies for reusing the same environment when encoding multiple lists of information—a necessity when considering the time consuming nature of creating memorable virtual environments. Chapter 5 demonstrates how mental representations of virtual environments can be decoded using functional magnetic resonance imaging (fMRI), setting the stage for an investigation into whether virtual encoding environments can be decoded during recall (the work conducted in Chapter 6).

Human memory is context-dependent, meaning that our internal and external environments becomes bound with memories that we form, thus affecting how well we learn and remember. This dissertation research used virtual reality (VR) to exploit this quirk to improve learning and retention. We found that learning two similar languages each in a distinct VR helped participants retain 92% of what they had learnt one-week later—but only if they had experienced VR as “real” environments. Moreover, using functional neuroimaging, this work identified brain patterns behind this effect, and provide novel and concrete evidence of contextually-supported memory recall and retention. Theoretical and technical implications of this work is discussed in terms of furtherance of neuroimaging and memory research, as well as possible clinical application and educational interventions. Additional discussions of the current state of the arts was also included for VR applications in neuroimaging, and applications of neuroimaging techniques in memory research.

The executive control of declarative memory is the goal-directed and volitional manipulation of representations from episodic and semantic memory systems and is thought to critically depend on the interactions between prefrontal cortex neural systems and the default mode network (Mitchell & Johnson, 2009; Spreng, Stevens, Chamberlain, Gilmore, & Schacter, 2010; Vincent, Kahn, Snyder, Raichle, & Buckner, 2008). The prefrontal cortex system itself is thought to be organized hierarchically along a rostrocaudal axis where increasingly anterior prefrontal cortex regions enact and represent more abstract executive control while more posterior frontal cortex regions exert progressively more concrete executive control interactions with motor systems (Badre & D’Esposito, 2009; Koechlin & Hyafil, 2007). The first chapter of this dissertation provides the background on the fronto-parietal control network and default mode network neural systems that govern the executive control of declarative memory (Andrews-Hanna, 2012; Fox et al., 2005; Power et al., 2011; Vincent et al., 2008). The second chapter discusses research focused upon further understanding how rostrolateral prefrontal cortex, the most anterior prefrontal region, interacts with neural systems involved in declarative memory, to enable episodic retrieval and analogical reasoning. The first section of Chapter 2 discusses an fMRI study that was performed to disentangle the role of rostrolateral prefrontal cortex in episodic memory retrieval and analogical reasoning and presents data showing that this brain region is domain-general in nature and functionally couples with domain-specific neural systems for goal-directed cognition (Westphal, Reggente, Ito, & Rissman, 2016). The second section of Chapter 2 presents data from a transcranial direct current stimulation study that causally influenced the stimulation levels of rostrolateral prefrontal cortex during the memory and reasoning tasks and presents evidence that the stimulation of the fronto-parietal control and default mode networks simultaneously improves episodic memory retrieval. The third chapter discusses research focused on further understanding the fronto-parietal control and default mode network neural systems. The first section of Chapter 3 discusses an fMRI study that assessed whole-brain functional connectivity during reasoning and memory tasks and shows that performance in episodic memory retrieval benefits from more whole-brain connectivity, in contrast to reasoning (Westphal, Wang, & Rissman, 2017). The second section of Chapter 3 presents data from a behavioral study that was inspired from work showing that major depressive disorder is associated with aberrant connectivity between the fronto-parietal and default mode networks (Hamilton, Chen, & Gotlib, 2013; Marchetti, Koster, Sonuga-Barke, & De Raedt, 2012), where the retrieval of negatively-valenced autobiographical memories interferes with an ensuing visual working memory task. The fourth chapter summarizes the main findings from each research study and provides some concluding thoughts on what this body of work has revealed about the neurocognitive control processes that support declarative memory.