The dopamine system, located primarily in the ventral tegmental area (VTA) and the substantia nigra, plays an important role in the reward-learning of both humans and mice. These dopamine neurons do not fire when a reward is received, but rather when there is a prediction error between what is expected and what occurs. Our understanding of dopamine neurons’ role in reward-learning and working memory has been established through short single cell recordings and pharmacological studies, but never through functional imaging of the relevant brain areas. This is a result of the depth of the ventral tegmentum and the inadequacies of the technology until now. This study carries out high-resolution functional imaging of ventral tegmentum dopamine neurons by inserting a cre-dependent GCaMP virus into the VTA of DAT-cre mice and then implanting a Gradient Index lens into the VTA to allow for two-photon microscopy. The VTA is imaged while completing a complex reward learning task in a virtual reality environment that simulates running down a T-stem maze. The raw data is then motion corrected using a unique algorithm that corrects for both fast jitter and slow drift movements, which allows single neurons to be recorded for extended periods of time. This functional imaging directly demonstrates that dopamine neurons follow the basic features of the traditional dopamine response, as the response shifts from reward to cue over the course of training and also diminishes as the reward becomes more expected. Another interesting finding is that single neurons are selective for specific visual cues during the task, but the selectivity of the neuron is not constant and can shift over time. These findings demonstrate the value of functional neuronal imaging in deep brain areas during complex behavior.

During slow-wave sleep, memory consolidation is driven by hippocampal replay of recent experiences. This replay triggers upstream reverberations with activity patterns that resemble those evoked by the original experience. In this study, we attempted to decode these reverberations in order to predict the degree to which subjects consolidated specific, recently-learned information. We trained subjects to associate sound cues with lateralized movements, then used those sounds to selectively bias memory reactivation during slow-wave sleep (SWS). A comparison of the electroencephalographic (EEG) signal evoked by left-associated and right-associated sounds revealed a laterally asymmetric, cue-dependent neural response during sleep. We detected significant differences in post-nap recall accuracy for cued and uncued stimuli, implicating this cue-dependent signal in the consolidation process. However, we detected no significant relationship between signal asymmetry and the effect of the cues on post-nap recall.

The current variety of treatment options for epilepsy leaves 30% of those who suffer from this chronic neurological disease without a cure. Therefore, this senior thesis project aims to uncover new insights about the brain structure that underlies susceptibility to epilepsy in hopes that a greater understanding of this underlying structure will catalyze the discovery of novel therapeutic methods which target these underlying differences in brain structure. To drive the discovery of new insights about underlying structure, this project addresses the following tension found in the literature: high frequency oscillations occur in both the brains of those with epilepsy and in the brains of those without epilepsy. Only when high frequency oscillations occur in the brains of those with epilepsy does the brain enter a state of unstable dynamics and seizure activity. This suggests that there is a difference in underlying structure between epileptic and non-epileptic brains, and this study uses computational modeling of neuronal firing to characterize these differences. First, based on a firing rate model, we find that within the phase space of the weight values, there is a band of stability from which one might predict the stability of a set of weights. Then, in the next two versions of the model, we add Hebbian plasticity and homeostatic plasticity. Only through the addition of Hebbian plasticity and homeostatic plasticity does high frequency oscillation, the manipulation described in our driving question, have a lasting effect on the weights. With the addition of a rate based Hebbian plasticity model to the base firing rate model, we find that weights can be perturbed from this band of stability through Hebbian plasticity. Adding a weight based homeostatic plasticity model to the base firing rate and Hebbian plasticity model then gives insight into the fact that having a target weight within a certain location with respect to the band of stability can rescue stability of a set of original weights from the destabilizing effects of Hebbian plasticity. Finally, we explore the effect of high frequency oscillation on various weight combinations within the phase space, and we find that certain weight combinations are projected to an unstable state through high frequency oscillation while other weight combinations remain at a stable state even in the face of high frequency oscillation. The unifying characteristic of those weights which remain stable in the face of high frequency oscillation remains an open question. However, in the process of investigating high frequency oscillations, it was found that weights on the edge of the band of stability are more robust to instability through Hebbian plasticity than weights on the band of stability that are further from the edge. These results suggest that the differential response to high frequency oscillation between epileptic and non-epileptic brains can be attributed at least in part to the location of weights with respect to the band of stability.

Recent work has suggested that mental health disorders may be understood in terms of dysfunction in the brain’s goal-directed Reinforcement-Learning (RL) architecture. More specifically, recent work has linked the compulsive aspect of drugs of abuse, and other disorders such as OCD and General Anxiety Disorder, to excessive use of automatic (model-free) over deliberative (model-based) action evaluation. In addition, a converse pathology of excess deliberation (model-based action evaluation) has been suggested to relate to other disorders (e.g. rumination in mood disorders). By assessing how symptoms of social anxiety disorder (SAD) predict model-based (vs model-free) learning in a socially framed RL task, we seek to investigate this hypothesis. In this experiment, we used a data-driven approach, employing large-scale online testing of 489 participants from a general population sample (Amazon Mechanical Turk). Social Anxiety Disorder (SAD) is then particularly interesting, and practical, to focus on both because of its prevalence in the Turk population, and due to it’s specific susceptibility to socially framed tasks. Therefore, subjects completed the Liebowitz Social Anxiety Scale (LSAS), and played 80 rounds of a competitive economic game, the Patent Race, against a computerized opponent. Previous research has captured human choices and neural responses on such tasks with a class of computational models referred to as the Experience Weighted Attraction (EWA) model. The EWA model nests two RL strategies, learning action values by a weighted combination of model-free reward sampling, vs. model-based learning (marginalizing) of the opponent’s move distribution. Estimating the parameters of the EWA model that best fit each subject’s choices, we extract parameters that characterize prior belief decay rate, belief updating, and learning rate. In accordance with our hypothesis, we found that self-reported social anxiety is related to a model parameter, delta, thought to reflect higher order belief learning selectively associated with increased use of model-based evaluation (P < .01; 6% increase in model-based learning per 1 SD increase in LSAS). These results are consistent with prior models of SAD that suggest the root of the disorder lies in excessive post-event processing that leads to paralysis in social situations, characterized by overthinking, heightened levels of mentalizing, and upward counterfactual thoughts. Yet, these results are unique in that they associate clinical symptomology with greater computational ability. By grounding the adverse symptoms of SAD in well-delineated neuro-computational mechanisms, these results illuminate the mental processes specific to SAD, and offer a rare example of enhanced function in disease.

One of the questions at the heart of neuroscience is how external experiences are translated into the language of neural spiking. Deemed the neural encoding problem, the relationship between a given stimulus and the resulting neural activity has long been probed by classical statistical methods. In this work, we seek to apply deep learning, an increasingly well-known and powerful way of modeling artificial neural networks, to study this problem in the early visual system. Deep learning is just beginning to be applied in such a way. We take it further by embracing a broad range of network architectures in order to highlight the ways in which the fundamental content of a neural signal changes from when it leaves the retina to when it reaches the brain, and how that content is molded by the different cells at each location. In attempting to model the fundamental way in which neurons communicate–their spike rate–we test our understanding of the underlying principles governing neural coding, explore benefits and drawbacks of past approaches, and achieve new levels of success at matching neural outputs.

Previous studies have shown that, in some domains, rats display significant cognitive similarity to both primates and humans, including the ability to hold information in working memory, and continuously update that information over the course of a task (Fassihi and Akrami, 2014). Rodents have also displayed systematic bias induced through stimulus statistics and trial history (Akrami et al, 2015). In fact, electrophysiological recordings from the posterior parietal cortex (PPC) show neurons carry significant amounts of information about the animal’s previous choice as well as recent sensory experience, thereby introducing the presence of a trial history signal (Akrami et al., 2015). However, the exact roles of other areas within the circuitry involved in the employment of a trial history signal are still unknown. This paper attempts to build upon these recent findings through the employment of anatomical and behavioral rodent paradigms. Anatomically, retrograde cholera toxin subunit B (CTB) was injected into the medial PPC (mPPC) to determine afferent projections. Histological analysis showed that the laterorostral part of the lateral posterior thalamus (LPLR) displayed the highest levels of fluorescent expression relative to other (non-sensory) areas. Therefore, it was selected as the upstream area with respect to the mPPC to study its role in working memory and trial history. A pilot study consisting of muscimol and saline infusions of the LPLR grossly deteriorated task performance, suggesting this area may be too sensitive to effectively inactivate using muscimol or liquid injection. The findings of this study may be useful in developing a better understanding of the circuitry underlying history signals as well as working memory as a whole. This knowledge may be applied to further the understanding of, and establishing diagnostic testing for learning disabilities (LD) and Alzheimer’s Disease (AD). However, further inactivation methods, with higher spatiotemporal resolution, are needed to help continue unraveling the circuitry underlying parametric working memory. Only then may its true clinical relevance be unveiled.

People cannot conduct more than a few, and usually only one, control-demanding tasks at a time. Although this limitation is a well-recognized feature of goal-directed behavior, the reason for this limitation has yet to be fully understood. At present, one of the more promising explanations of this phenomenon is that the execution of multiple tasks is limited by cross-talk between overlapping task processing pathways (Allport, 1980; Feng, Schwemmer, Gershman, & Cohen, 2014; Meyer & Kieras, 1997; Navon & Gopher, 1979; Salvucci & Taatgen, 2011). The current work seeks to test the hypothesis that the emergence of overlapping task representations between tasks based on task similarities in the environment promotes learning efficiency, but impairs multitasking performance (Musslick et al., In Press, 2016b). I will present an experiment in which subjects learned multiple tasks that relied to different degrees on similar task-features. After learning all tasks, multitasking performance was assessed. The results of the experiment suggest that new tasks can be learned at a faster rate if they share features with previously acquired tasks, although early learning may be impaired. Our results yield higher interference for multitasking conditions in which the tasks relied on features shared with other tasks as opposed to independent tasks, although this did not significantly change multitasking accuracy. These antagonistic performance effects of learning rate and multitasking interference may indicate a computational tradeoff between learning efficiency through the use of shared representations and multitasking performance.

Neurotechnology is expanding rapidly, and it seems to be a matter of time before many products are widely available. One of the first technologies on the consumer market is transcranial direct current stimulation (tDCS). Electrical stimulation has a long history. tDCS has recently garnered research, clinical, and consumer interest. It employs 1-2mA current, thought to induce electric fields that alter cortical excitability. While tDCS has had dubious results on a myriad of clinical and cognitively enhancing indications, it is advertised for improving mood, motor skills, working memory, and more. A tDCS community has been inspired by that research, emboldened by the simple and inexpensive design of tDCS. These individuals build do-it-yourself (DIY) products and buy direct-to-consumer (DTC) devices, which the Food and Drug Administration (FDA) recently classified as lifestyle products rather than medical devices. Sociological research shows that these individuals want authorities to guide them through the research. The FDA and many researchers have noted the need for thoughtful policy proposals. This thesis proposes using meta-analysis to evaluate tDCS devices and inform consumers. It explores three clinical analogues of oft-advertised indications: craving, chronic stroke motor improvement, and depression. All three displayed small-to-moderate effects compared to sham (Hedges’ g craving = 0.340; g depression = 0.531; g stroke = 0.486). A case study on Focus, one of the largest tDCS companies, revealed that in working memory tasks, Focus’ device compares unfavorably to the tDCS standard and other interventions. Based on this proof of concept, this thesis explores the subsequent policy and ethical considerations—making recommendations for how to use a data-driven approach to constructively engage with the tDCS industry and community. This model could be constructively used to inform consumers about technologies that are either scientifically or ethically dubious.

Biasing memory reactivation with associative cues during non-rapid eye movement (NREM) sleep has been shown to increase recall and performance on declarative memory tasks. The precise distribution of these cues within NREM, however, particularly with regards to intermediate sleep stages like stage 2, has not been studied previously. Through the use of a real- time sleep spindle detection script, we look into the possibility of controlling cue timing to optimize memory reactivation, manipulating cues to be presented either within the refractory period of a spindle (“early” cues) or outside of it (“late” cues). Using an offline spindle detector as the golden standard, the online script showed reasonable levels of precision and recall in spindle classification (~0.55, ~0.40, respectively) and high efficiency in sound presentation, with a mean processing time on an order of magnitude of -4 after detection. As hypothesized, late cues resulted in improved recall performance over early cues on a sound-picture-location declarative memory task. However, interestingly, cues that appeared just on the limit of a spindle’s refractory window (~3-6 seconds) showed best recall, and those before or after this period do not elicit significantly enhanced memory performance. It is likely that these cues occur during a “spindle-rich period,” in which spindles occur continuously within 3-6 sec intervals of one another. Thus, overall, from these results we can conclude that 1) spindle detection is possible on a real-time scale; 2) cues that occur outside the refractory period of spindles are better recalled than those appearing within the period; and 3) there is a 3-6 second window right after a spindle occurrence in which a cue can occur for optimal memory reactivation and performance, likely during a spindle-rich period of time.

This paper presents data from an experiment on visual attention. The purpose of the experiment was to test a novel experimental paradigm for research into the relationship between visual attention and awareness in hemispatial neglect patients. The experiment was performed on normal volunteers; followup experiments would be performed on clinical patients with hemispatial neglect. I provide an overview of established literature regarding the neurological disorder known as hemispatial neglect, and how the disorder can be informed by Michael Graziano’s Attention Schema Theory of consciousness. By building on previous studies conducted on the attention schema and awareness, I propose a novel task paradigm for use in visual behavioral experiments with hemispatial neglect patients. This experimental design utilizes the position-based disparity in awareness seen in neglect patients to study both the effects of awareness on attention, and the nature of the underlying mechanism disrupted in hemispatial neglect. This study explores the implementation of this model in behavioral experiments conducted on healthy human subjects, and the implications of its use in the future for the study of neglect. Through the findings of previous and current research discussed in this paper, I provide new methods and context for the further understanding of the brain’s mechanism of controlling attention and awareness.

Cognitive control is a set of cognitive functions that guide behavior in service of a goal. Though decisions about goals and actions benefit from the use of cognitive control, cognitive control is limited and therefore must be allocated accordingly. Research suggests that all else equal, people prefer to avoid exerting cognitive control, but there are individual differences in the extent of these preferences. These preferences about cognitive control reflect cost-benefit analysis in the brain, which may have consequences on specific decision-making strategies that people implement. In our present study, we first attempted to quantify cost of control through a novel variant of the patch-foraging task originally developed by Constantino & Daw (2015). In our version, we added a cognitive load in order to examine the extent to which cognitive demand discounts reward. Empirical results revealed that the main Stroop manipulation did not reliably shift behavior across blocks in the current design, and the effect was trending in the opposite direction as our hypothesis. This inconsistent effect of Stroop incongruency suggests that either our present cognitive demand manipulation was not costly enough for our subjects, or that there was an additional factor that shifted the baseline effect of incongruency upwards, in which case individual differences around the mean may still be meaningful. To test this second possibility, we tested whether the variation in the incongruency effect had any relationship with the horizons task developed by Wilson et al. (2014), which quantifies individuals’ use of more (directed) or less (random) deliberative strategies. We hypothesized that individuals with lower costs of cognitive control will employ more deliberative decision strategies than those with higher costs. Although we found our results to be trending in the same direction as our predictions, the effects were insignificant. These two results taken together suggest the current foraging design was unable to index cognitive control, and thus it is still unknown whether cost of control, when effectively measured, predicts the balance of exploration strategies.

Over 800,000 individuals in the United States suffer from a stroke or a traumatic brain injury each year. These incidents often leave patients with lasting neurological and neuropsychiatric disorders, which affect their brain function for the rest of their lives. Prosthetic devices can be used to replace some sensory and motor defects. However, a prosthetic for cognitive disorders such as hemispatial neglect has yet to be developed. Previous cognitive prosthetic attempts have only interacted with a small portion of the brain, and only help patients regain function on behaviors that were “prerecorded”. Here, we begin development of an Adaptive Cognitive Prosthetic (ACP) that uses an adaptive learning algorithm paired with multi-site electrical microsimulation and recording techniques to record brain activity from normally functioning regions of the brain, transform this signal to mimic the lost function, and then stimulate a healthy brain region with this signal. This system has two advantages. First, it can use a larger microelectrode array to control many neurons at once, rehabilitating patients more effectively and flexibly. Second, it can learn over time, adapting the prosthetic to the changing needs and environmental conditions of the patient. There are two phases to this project. In the first phase, we develop large-scale electrical microstimulation techniques on mice. In particular, we develop the learning algorithms supporting the ACP. This allows the ACP to guide neural activity towards a target response, often reaching that target within 10 minutes of searching for the correct stimulation pattern. In the second phase of the project, we explore whether control over brain activity can translate to control over behavior. To this end, we use both electrical and visual stimuli to stimulate V1 neural activity in anesthetized mice. Importantly, we find the responses to both electrical and visual stimulation are similar. This similarity is crucial, as the prosthetic must be able to produce neural responses that are similar to neural activity during normal function in order to replace this function. These results will form the basis for behavioral experiments that will test the ACP’s ability to replicate perception.

The cerebellum is the most frequently observed site of structural abnormality in individuals with autism spectrum disorder (ASD), but its role in the etiology of ASD is unknown. We hypothesized that the cerebellum guides the maturation of distal neocortical regions during sensitive periods of development. The developmental diaschisis hypothesis predicts that cerebellar disruptions can impair the refinement of neocortical circuits, leading to ASD-related pathologies. Using the Purkinje cell-specific L7-tuberous sclerosis 1 (Tsc1) heterozygous mice as a model of pan-cerebellar perturbation, we found increased spine density on apical and basal dendrites of layer II/III pyramidal neurons in the medial prefrontal cortex (mPFC). Previous work on the Tsc1 model demonstrated that heterozygous Tsc1 mice display ASD-like behavioral deficits including social impairment and repetitive behavior. We explored whether region-specific cerebellar perturbations also alter social behavior and spine plasticity in functionally correlated neocortical regions. Designer Receptors Exclusively Activated by Designer Drugs (DREADDs) were expressed in cerebellar lobule VI or crus I via region-specific injections of AAV8-hSyn-hM4D(Gi)-mCherry to inhibit activity of molecular layer interneurons throughout development (postnatal day 21-60). In the three-chamber sociability assay, mice with DREADDs expressed in lobule VI displayed significantly decreased social preference compared to controls injected with saline. We found no significant difference in spine density and morphology of layer II/III pyramidal neurons in the mPFC or in layer II/III and V pyramidal neurons in the anterior cingulate cortex between the DREADDs and saline-control groups. This study provides experimental evidence and suggests possible synaptic mechanisms for the role of the cerebellum in the development of flexible and social behavior and in the maturation of neocortical circuits.

To engage in goal-directed behavior, we must be able to predict the consequences of our actions. When we drop a glass or strike a match, for example, there is a specific outcome we know to expect. Though the generation of predictions based on such long-term semantic memories remains to be characterized, recent research suggests that predictions based on specific episodes of the past may be facilitated by pattern completion in the hippocampus (Hindy et al., 2016). Hypothesizing that long-term memory prediction continues to be supported by the hippocampus, we collected high-resolution fMRI data while participants selected among familiar actions that changed the state of one object into another. In a ‘novel’ condition representing shorter-term episodic memory, participants learned actions that changed objects in arbitrary ways, requiring predictions to be based on episodes experienced recently, three days before fMRI scan. In a ‘known’ condition representing long-term semantic knowledge, participants could predict the outcomes of selected actions by utilizing their pre-existing knowledge. In both conditions, responses in early visual cortex were attenuated when an outcome could be predicted from an action, and this predictive attenuation was related to background connectivity with the hippocampus. Further segmentation of the hippocampus itself provided evidence for functional distinction along the long axis. Namely, prediction of novel cue-action outcomes in visual cortex was related to the anterior hippocampus, and prediction of known cue-action outcomes was related to posterior hippocampus. These findings indicate that the hippocampus mediates visual prediction based not only on specific episodes, but on long-term semantic knowledge as well. The generation of predictions utilizing these two types of memory, however, may be facilitated by different mechanisms.

In reward learning, the difference between the expected value of an outcome and the observed outcome is the prediction error. Reinforcement learning models use a learning rate parameter to discount the prediction error when the model updates the expected value of a stimulus. Previous work has demonstrated that learning agents adapt their learning rates according to the uncertainty of the context. In this thesis, I analyzed data from a task in which participants learned reward values in two risk contexts with different degrees of uncertainty. I computed and compared empirical learning rates, learning rates fitted by reinforcement learning models using conventional log likelihood maximization, and learning rates estimated by hierarchical Bayesian modeling. The results from all three analyses suggested that participants had higher learning rates in the low-risk context. Hierarchical Bayesian modeling estimated learning rates more robustly than the other two methods. I used hierarchical Bayesian modeling to examine participant learning rates, which decayed over time within an uncertainty context. Finally, I demonstrated that positive prediction errors were associated with higher learning rates in the low-risk context but not the high-risk context. This novel finding suggests that high uncertainty might interfere with optimistic over-attention to positive outcomes.

Traumatic brain injury (TBI) affects over 57 million people globally each year, of whom 1.4 million live in the United States. There are 475,000 new cases of pediatric traumatic brain injury per year, accounting for 3,000 to 4,000 deaths. Traumatic brain injury (TBI) differs from ischemic stroke and other non-mechanical injuries in that head trauma exposes cells of the central nervous system (CNS) to shear force, tensile stress, and compression forces. In addition to physical disruption of neural structures, these forces induce damaging activation of cellular stress and inflammation pathways. The resulting cellular processes following a traumatic brain injury may damage DNA, including the telomeres. Telomeres are the end chromosomal structures comprised of tandem repeating base pairs (of nucleotide sequence TTAGGG) that protect genetic information. Due to the end replication problem, in the absence of telomeres, the chromosomal ends shorten as DNA replicates. The presence of the telomere buffers this effect. In some cell lineages, telomerase maintains telomere length during replication. In most somatic cells, however, telomere attrition occurs naturally due to aging, oxidative stress, and other mechanisms. Further, telomerase activity is implicated in neurogenesis and may affect how neural and glial cells survive. Cells with active telomerase are more resistant to cellular apoptosis. Recent literature suggests that telomere attrition may be accelerated by TBI and TBI may induce telomerase activity. Previous studies show that telomerase activity was induced in mice hippocampi following kainate-induced seizures and telomere lengths shortened in some rats as a result of a mild concussive TBI, but experimental conditions did not permit a definitive statement as to the link between the level of trauma and telomere length. It is not clear how a moderate traumatic brain injury affects telomerase activity and telomere lengths. To address this gap, we sought to measure telomerase activity and telomere lengths in cells of the CNS both 24 hours and 14 days following a moderate Controlled Cortical Impact (CCI) injury in adult male mice. We found changes in telomerase activity in the ipsilateral cortex, thalamus, and hippocampus 24 hours following injury, and telomerase activity remained present in these regions 14 days following injury. Telomerase activity in some tissues were statistically significant when compared to controls. Additionally, we found significant telomere length changes in the cells of the CNS 14 days following TBI. Telomere elongation occurred in the ipsilateral cortex, telomere maintenance occurred in the ipsilateral hippocampus, and telomere attrition occurred in the ipsilateral thalamus. Correlation between a moderate traumatic brain injury and telomere length mediated by telomerase activity 1 day after a moderate CCI injury may suggest a possible molecular mechanism by which telomeres change after a traumatic brain injury, providing insight for possible future medical treatment and intervention. Future studies should investigate the types of cells expressing telomerase and changes in telomere lengths to better understand the implications that telomere length and telomerase activity have in neurogenesis and to elucidate the cellular mechanism by which telomere changes occur following a traumatic brain injury.

Attention deficit/hyperactivity disorder (ADHD) is a strikingly prevalent neurodevelopmental disorder amongst children and adolescent populations. More so, the disorder is becoming increasingly predominant in adults, nevertheless the scientific literature surrounding ADHD remains divided on crucial characteristics: the age of onset, the diagnostic criteria, the treatment methods and especially the causes of the disorder. ADHD is primarily regarded as atypical levels of inattention, impulsivity and hyperactivity, with management techniques ranging from pharmaceutical interventions to behavioral therapies. However, no true cognitive assessment exists to gauge the disorder appropriately, which lends itself to ambiguity and an excessive potential for incorrect diagnoses. This thesis aims to provide a cohesive analysis of the contemporary research surrounding ADHD in efforts to structure a more direct and clinically useful diagnostic technique founded on a principal theoretical framework of attention. Ultimately, this perspective should frame latent research to discover suitable and effective cures for ADHD.

In some species, vocalization represents a vital asset for survival and forging connections with other individuals. Humans and marmosets both participate in a vocal ritual known as turn taking, thereby allowing the marmoset to serve as a compelling model system for human speech. The coupled oscillator model, proposed in a previous study, uses inter-call intervals to explain how marmosets time their vocalizations to avoid overlap in paired scenarios. In this study, I turn my attention to an understudied parameter of marmoset vocalizations; call duration, in the hope of simultaneously rationalizing a confounding result present in developmental data, and illuminating the role of call duration itself within the context of this oscillator model. In my statistical analysis of “alone” and “paired” marmoset data collected from a previous study, I uncovered a significant, positive relationship between initial and response call duration. I also designed a playback experiment to isolate this relationship specifically, but no response vocalizations from the subjects emerged. Despite this, my experimental design provides a template for future studies aiming to investigate the relationship between initial and response call duration. Additionally, the result I uncovered in my data analysis suggests that marmosets use call duration in a similar manner that inter-call intervals are used in the coupled oscillator model: to prevent vocal overlap. Since eschewing interruptions during a vocal exchange is a crucial part of human speech, this study more generally provides insight into the intricacies of human vocalization timing.

The cortex is a very important brain structure which coordinates sensory perception, cognition, and motion in mammals. Crucial to the structure of the cortex is the fact that all areas are made up of the same repeating motif, known as the canonical cortical circuit. One theory about the function of the circuit is that its fundamental job is that of sequence recognition and prediction. This thesis takes a multi-pronged approach to advancing the understanding of cortical sequence processing. It describes two experiments done in V1 in mouse, both of which investigate how layer 2/3 cells respond to violations in an expected order of images. These experiments found that V1 is able to learn the statistics of the experiment over time and adjust the strength of its predictions accordingly. In addition to these experiments, the thesis describes the results two projects of data analysis on experiments which had previously been conducted by someone else. The first data analysis investigates how the duration for which each image is presented to a mouse affects a variety of neural responses. The second data analysis attempts to discover how the hippocampus responds to violations in an expected image order, and concludes that no response could be found. Finally, the thesis contains a literature review which describes the relationship between stimulus specific adaptation and the mismatch negativity in the auditory cortex, two phenomena related to sequence processing. Overall, by taking a multi-pronged approach, this thesis seeks to advance our understanding of how the canonical cortical circuit processes sequences, as well as shed light on the areas which need to be investigated further.