APRIL 23–26 • 2022

CNS 2022 Annual Meeting | Data Blitz Session Schedule

A Data Blitz is a series of 5-minute talks, each covering just a bite-sized bit of research. It will offer a fast-paced overview of some of the most exciting research presented at this year's poster sessions.

During the abstract submissions process first authors had the option to indicate whether they’d like their abstract to be considered for the Data Blitz Session. Selected abstracts have been scheduled into the Data Blitz session in addition to a poster session. Talks will be given by the abstract first author — a faculty member or student at any level. 

Each DataBlitz session includes 15-5 minute talks






Data Blitz Session 1 Saturday, April 23 2:00 - 3:30 pm Grand Ballroom A Marian Berryhill
Data Blitz Session 2 Saturday, April 23 2:00 - 3:30 pm Grand Ballroom B/C Chelsea Helion
Data Blitz Session 3 Saturday, April 23 2:00 - 3:30 pm Bayview Room Elizabeth Norton
Data Blitz Session 4 Saturday, April 23 2:00 - 3:30 pm Seacliff Room Vishnu Murty

Data Blitz 2022 Schedule


Add to Calendar
04/23/2022 2:00 PM
04/23/2022 3:30 PM
Data Blitz Session 1
Data Blitz Session 1 will be held in person at the CNS 2022 Annual Meeting in San Francisco at the Hyatt Regency San Francisco Hotel. Located at 5 Embarcadero Ctr, San Francisco, CA 94111 in the Grand Ballroom A
Grand Ballroom A

CNS 2022 | Data Blitz Session 1

Talk 1: Just a phase? Probing the causal role of neural oscillations in human behaviour.

Matteo Vinao - Carl, Imperial College London

For over a decade, findings from electrophysiological studies suggest a fundamental role of oscillatory phase in shaping human behaviour. However, empirical evidence of this relationship is correlative and not causal. To address this, we developed a phase-locking strategy to causally probe the role of oscillatory phase in visual attention. In this approach, visual cues were synchronised in real-time to the instantaneous phase of spontaneous neural oscillations recorded using electroencephalography (EEG). Our findings from two experiments did not support a causal link between the phase of occipital alpha (10Hz) or frontal-midline theta (5Hz) and attentional performance. However, consistent with previous studies, performance was strongly correlated with alpha and theta phase when the phase was computed offline. Further analysis revealed that this spurious correlation is an artefact resulting from the evoked potential (EP), which confounds gold-standard phase computation approaches such as wavelet convolution and the Hilbert transform. We discuss the implications of this finding for EEG research and present a new method for uncovering true phase-behaviour relationships in electrophysiological data.

Talk 2: Distinct Contributions of the Cerebellum and Basal Ganglia to Arithmetic Procedures

William Saban, University of California, Berkeley

Humans exhibit complex mathematical skills, often attributed to the exceptionally large neocortex. Using a neuropsychological approach, we report that degeneration within two subcortical structures, the basal ganglia and cerebellum, impairs performance in symbolic arithmetic. Moreover, we identify distinct computational impairments in individuals with Parkinson's disease (PD) or cerebellar degeneration (CD). The CD group exhibited a disproportionate cost when arithmetic sum increased, suggesting that the cerebellum is critical for iterative procedures required for calculations. The PD group exhibited a disproportionate cost for equations with an increasing number of addends, suggesting that the basal ganglia are critical for the coordination of multiple cognitive operations. In Experiment 2, the two patient groups exhibited intact practice gains for repeated equations at odds with an alternative hypothesis that these impairments were related to memory retrieval. Overall, the results provide a novel demonstration of the contribution of subcortical structures to the computations required for complex cognition.

Talk 3: Investigating the role of locus coeruleus degeneration in attentional control in prodromal Parkinson's disease

Sophie Sun, McGill University

Noradrenergic neurons in the locus coeruleus are thought to degenerate early in Parkinson's disease (PD), possibly as early as the prodromal stage. Noradrenaline is also known to play an important role in supporting selective attention and response inhibition. Whether specific deficits related to degeneration in the locus coeruleus can be detected as early as prodromal Parkinson's disease remains unknown. To address this, we are measuring attentional control using a visual oddball task. Degeneration in the locus coeruleus is measured using neuromelanin and diffusion-weighted MRI-derived measures of microstructural integrity (signal intensity, fractional anisotropy, and mean diffusivity). Participants include prodromal Parkinson's patients, defined as the presence of REM sleep behaviour disorder (current n=17, MoCA=26), Parkinson's patients (n=52, MoCA=26), and older controls (n=31). Preliminary behavioural analyses indicate that controls have better attentional control (accuracy=0.95, SEM=0.02) compared to prodromal patients (accuracy=0.90, SEM=0.03, p<0.01) and PD patients (accuracy=0.90, SEM=0.03, p<0.01). Response times (in milliseconds) are similar for controls (M=494, SEM=22) and prodromal patients (M=488, SEM=19, p=0.47), whereas PD patients are slower than both groups (M=546, SEM=39, p<.001 for both). Ongoing analyses aim to relate individual differences in attentional control to MRI-derived measurements of degeneration in the locus coeruleus. Identifying early brain-behaviour relationships will help us understand the mechanisms underlying the cognitive profile of Parkinson's patients and allow us to distinguish the specific contribution of degeneration in the locus coeruleus from the more generalized degeneration that occurs as the disease progresses.

Talk 4: Brain activation during goal-related selection is predicted by individual differences in prefrontal GABA+ concentration

Boman Groff, University of Colorado Boulder

Prior work suggests that GABAergic mechanisms, which control inhibitory neural dynamics, can influence how quickly a choice between multiple task-relevant options (goal-related selection) occurs. For instance, administering the GABA agonist midazolam increases the ability to select amongst competing task-relevant options in a verb generation task, most notably when such selection is difficult (Snyder et al., 2010). Moreover, individuals with a relatively higher concentration of GABA+ compared to glutamate/glutamine (Glx) in lateral prefrontal cortex (LFPC) show less of a decrement in selection abilities under difficult conditions (de la Vega et al., 2014). While this prior work focused on associations between GABAergic mechanisms and behavior, the present study tested whether individual differences in neurotransmitter concentration in LPFC were associated with brain activation during a verb generation task. In a sample of youth and emerging adults (N=64, 32 males/32 females, age range: 16-25 years), PRESS and MEGAPRESS spectroscopy sequences determined resting GABA+ and Glx concentrations in two separate voxels (dorsal LPFC, ventral LPFC). Participants then underwent functional magnetic resonance imaging while performing a verb generation task that manipulated the difficulty (high, low) of goal-related selection. Individuals who had higher resting concentrations of GABA+ in ventral LPFC exhibited less of a difference in prefrontal activation for hard vs. easy conditions. This finding is similar to what we previously observed in a group of middle-aged women who were in a parallel study. Hence, the current findings suggest that the association between GABA+ and goal-related selection is robust and consistent across different age groups.

Talk 5: A Novel Gamma- and Music-Based Intervention for Alzheimer's Disease and Dementia-Related Disorders

Parker Tichko, Northeastern University

Alzheimer's disease (AD) and other dementia-related disorders are thought to arise, in part, from disrupted gamma-band neural activity (30 - 100 Hz) and phase-amplitude coupling (PAC) between gamma and slower bands of brain activity. Recent work with non-human-animal models of AD suggests that non-invasive auditory and visual gamma-frequency stimulation can remediate multiple pathophysiologies of dementia and improve cognitive functioning. Here, we present a non-invasive, music-based intervention for treating dementia-related disorders, called SynchronyGamma, that aims to induce PAC by delivering gamma visual stimulation during natural music listening. SynchronyGamma is a light controller that synchronizes LED lights to the rhythm of music in the delta (0.5 - 4 Hz) and theta (4 - 8 Hz) frequency bands, while coupling gamma-frequency visual stimulation to the frequencies of the musical rhythm. In a sample of young adults (N = 32) and older adults without dementia (N = 32), we tested whether SynchronyGamma drives gamma neural activity, induces PAC between delta, theta, and gamma brain activity, and improves visual working memory. Relative to a control intervention featuring music-listening and synchronized lights without visual gamma stimulation, SynchronyGamma was found to drive scalp-recorded gamma activity and induce stronger PAC between delta-gamma and theta-gamma brain activity. Moreover, older participants performed more accurately on a visual working memory task compared to their control counterparts, although this improvement did not reach statistical significance. Together, these findings suggest that a music-based intervention paired with visual gamma stimulation is effective at driving gamma neural activity, inducing PAC, and may improve cognitive functioning.

Talk 6: Age modulates the oscillatory neural dynamics serving fluid intelligence

Samantha Penhale, Boys Town National Research Hospital

Fluid intelligence (Gf) encompasses logical reasoning and novel problem-solving skills and is often assessed using abstract reasoning tasks like Raven's progressive matrices. Previous work has reported an age-related decline in fluid intelligence capabilities, and although many studies have sought to identify the underlying mechanisms, our understanding of the critical brain regions and dynamics remains largely incomplete. In this study, we utilized magnetoencephalography (MEG) to investigate 78 individuals, ages 20 - 65 years, as they completed an abstract reasoning task. MEG data was transformed into the time-frequency domain and the resulting neural oscillations were imaged using a beamformer. We found worsening behavioral performance with age, including prolonged reaction times and reduced accuracy. MEG analyses indicated robust oscillations in the theta, alpha/beta, and gamma range underlying abstract reasoning. Our primary results included stronger alpha/beta oscillations with increasing age in the right superior parietal cortex, a key area previously found to be implicated in intelligence and reasoning. In addition, whole brain correlation analysis using age revealed a similar relationship in bilateral parietal and left frontal cortices, both regions thought to be involved in Gf. Follow-up connectivity analyses using these regions as seeds revealed increasing parieto-frontal connectivity with age. Importantly, our findings are consistent with the parieto-frontal integration theory of intelligence (P-FIT). These results further suggest that as people age, there may be a compensatory effect that is spectrally specific, such that older people exhibit stronger alpha/beta oscillations across the parieto-frontal network to solve abstract reasoning tasks.

Talk 7: Both aperiodic activity and slow-oscillatory power increase in patients with treatment-resistant depression after receiv

Quirine van Engen, UCSD

Major Depressive Disorder (MDD) is one of the most prevalent mental disorders in the world, and is associated with long-lasting impairments to executive functioning. For otherwise treatment-resistant MDD, electroconvulsive therapy (ECT) remains one of the most widely-used and efficacious interventions. Despite its popularity, ECT's mechanism of action remains unknown. Although ECT has been associated with increases in the amplitude of low-frequency activity in the human electroencephalogram (EEG), commonly referred to as 'EEG slowing,' how this change in the EEG relates to clinical improvement has yet to be resolved. Up to this point, increases in slow-frequency power have been assumed to indicate increases in slow oscillations, without consideration for the contribution of non-oscillatory, aperiodic activity, which, importantly, has a different physiological mechanism. The objective of this study was to evaluate the potential role of aperiodic activity in EEG responses to ECT. Twenty-three patients with treatment-resistant depression underwent a course of ECT and EEG data was recorded at baseline and after completing treatment. Using spectral parameterization methods, we show that the EEG aperiodic exponent significantly increases with successive ECT treatment, evident in a visible spectral 'steepening,' in addition to increases is oscillatory power in the delta (1-4 Hz) and theta (4-7 Hz) bands. Increases in aperiodic exponent, like those seen here, are related to increased inhibitory activity according to computational models of excitation-inhibition balance. This result aligns with long-standing theories of cortical inhibition in depression, indicating that ECT might ameliorate depressive symptoms by restoring healthy levels of inhibition in frontal cortices.

Talk 8: Altered verbal working memory neural dynamics in preclinical and prodromal Alzheimer's disease

Amy Proskovec, University of Texas Southwestern Medical Center

Background: In healthy adults, neural oscillatory dynamics within a predominantly left-lateralized network of brain regions underlies verbal working memory (VWM) performance, but how the preclinical (subjective memory complaint [SMC]) and prodromal (amnestic mild cognitive impairment [aMCI]) stages of Alzheimer's disease (AD) impact these dynamics is not well characterized. We utilized magnetoencephalographic (MEG) brain imaging to investigate the effects of SMC and aMCI on the neural oscillations serving specific phases (i.e., encoding, maintenance) of VWM.

Methods: Eighty-four adults (39 cognitively healthy [CH], 24 SMC, 21 aMCI, 44 female, M age = 63.1) completed a VWM task during MEG. All MEG data underwent standard preprocessing, were transformed into the time-frequency domain, and significant oscillatory responses relative to baseline were imaged using a beamformer. To determine the effect of group (CH, SMC, aMCI), ANCOVAs were performed on the resulting encoding and maintenance whole-brain maps with age as a covariate, and follow-up t-tests were performed.

Results: Across groups, decreases in alpha-beta (9-16 Hz) activity were seen in left fronto-temporal regions throughout encoding and maintenance. Significant group differences emerged in the right anterior cingulate, inferior frontal, and superior temporal gyri during encoding (all p's < .05, corrected). Both SMC and, to a lesser degree, aMCI individuals exhibited increases in theta (4-7 Hz) activity within these regions, while these oscillatory responses were absent in CH individuals.

Conclusions: Adults with preclinical and prodromal AD recruited additional right fronto-temporal regions during VWM performance. Our results offer supporting evidence for the compensation-related utilization of neural circuits hypothesis (CRUNCH).

Talk 9: Pediatric intracranial EEG recordings reveal age differences in occipital alpha oscillations and aperiodic activity

Qin Yin, Wayne State University

Intracranial EEG recordings provide invaluable information about the neurophysiological mechanisms of cognitive development. Electrophysiological data consist of oscillatory and aperiodic activity, however, and developmental differences in these two signals are largely unknown. We investigated age differences in neural oscillations and aperiodic activity using data from 154 electrodes implanted in the occipital cortex of 25 children and adolescents (6.20-20.50 years) who were undergoing direct cortical monitoring for seizure management. Subjects studied pictures of scenes in preparation for a memory recognition test. The power spectrum was calculated using Welch's method for each 3-s scene encoding trial and 0.45-s pre-trial baseline. Power spectra were averaged over trials (separately for subsequent hit, subsequent miss, and baseline period) then entered into the FOOOF algorithm to estimate neural oscillations (central frequency, power, bandwidth) and aperiodic activity (offset, exponent) from 1-30 Hz. Linear mixed-effects models with age and condition as fixed factors were used to test between-subjects' effects. The detected dominant oscillations overlapped with alpha range in all conditions (subsequent hit: 4.39-11.15 Hz; subsequent miss: 4.44-11.68 Hz; baseline: 4.76-11.40 Hz). Critically, in all conditions, age-related increases were observed in the oscillatory frequency (F(1,456)=14.02, p<.001) and bandwidth (F(1,456)=14.02, p<.001), but not in power (F(1,456)=1.57, p=.21). Furthermore, in all conditions, modest age-related decreases were observed in the offset (F(1,456)=4.12, p<.05) and exponent (F(1,456)=4.20, p<.05) of the aperiodic activity. These results highlight the age differences in narrow-band oscillation frequency and broad-band power, suggesting the importance of isolating oscillatory and aperiodic activity when studying the development of neurophysiological mechanisms of cognition.

Talk 10: Are Rare Stimuli that Produce a P3b Component Preferentially Encoded in Working Memory?

Carlos Carrasco, University of California, Davis

The P3b ERP component is hypothesized to reflect context updating, where a mental model of the environment is evaluated and amended when rare stimuli are encountered. This updating may lead to a higher fidelity representation for rare, P3b-eliciting items held in working memory. To test this hypothesis, we designed an oddball task in which each stimulus was a disc presented at one of 8 locations around a circle. The locations were divided into two groups-one rare and one frequent-and participants responded on each trial to indicate the group. Participants were also occasionally probed to report the exact location of the preceding disk with a mouse response. As expected, stimuli in the rare group produced a larger P3b than stimuli in the frequent group. Furthermore, when participants were probed to report the location of a disc, the report was more precise for rare items than for frequent items, indicating better working memory encoding. Surprisingly, preliminary analyses of the ERPs using decoding methods showed no evidence of greater location decoding accuracy for rare stimuli compared to frequent stimuli. Additional analyses will be conducted to attempt to resolve the discrepancy between the behavioral results and the ERP decoding results. However, the behavioral results provide clear evidence that rare, P3b-eliciting stimuli are encoded more precisely in working memory.

Talk 11: The Dual-System Model as Indexed by Executive Functions, Reward Sensitivity, and White Matter Tractography: Impacts of Age and Sex

Vanessa R. Alschuler, University of Minnesota

The dual-systems model posits distinct developmental trajectories in reward (i.e., quadratic) and executive (i.e., linear and/or inverse) systems through adolescence and early adulthood, resulting in elevated reward sensitivity relative to executive control in adolescence. Whether these putative developmental trajectories differ by sex remains equivocal. Accordingly, this cross-sectional analysis aimed to investigate whether sex differences in age-related changes of reward and executive function are compatible with the dual-systems model. A sample of age-matched males (n=75; mean age=17.77 ±3.70 years, range=11.38-24.93) and females (n=90; mean age=18.45±4.28 years, range=11.25-25.70) completed questionnaire measures of reward sensitivity (Behavioral Activation System; BAS), and assessments of verbal working memory (Digit Span; DS), nonverbal working memory (CANTAB Spatial Working Memory; SWM), and delay discounting (DD) as part of a longitudinal neuroimaging study. Curve-fitting procedures compared whether linear, quadratic, or inverse effects best approximated age-related patterns of development for each measure in males and females. Males showed a quadratic age effect for reward sensitivity (BAS Total t=-2.5, p=0.017) and linear age effects for working memory (DS t=3.8, p<0.001; SWM t=-4.4, p<0.001) and discounting (DD t=3.7, p<0.001), while females showed inverse age effects across all measures (BAS Total t=-1.8, p=0.077; DS t=-3.7, p<0.001; SWM t=-2.0, p=0.046; DD t=-0.4, p=0.69). These findings imply that the dual-system framework may be more relevant for males and that reward sensitivity develops differently in females. Further analyses are currently underway to test this model in relation to white matter development using probabilistic tractography of accumbofrontal and frontodorsostriatal tracts. Implications for adolescent risk-taking will be discussed.

Talk 12: Top-down control of multi-item working memory representations

DanielPacheco, Ruhr University Bochum

Visual working memory depends on both material-specific posterior brain areas that support the representation of the to-be-maintained stimulus features and executive control areas in the prefrontal cortex (PFC). These two areas putatively support bottom-up (storage) vs. top-down (control) functions that rely on oscillations in the gamma and beta frequency bands, respectively. A previous intracranial EEG (iEEG) study combined representational similarity analysis (RSA) and deep neural networks (DNNs) as a model of the processing hierarchy along the ventral visual stream (VVS) and showed that maintenance of natural images relied on both high-level visual and semantic information (Liu et al., PNAS 2020). How selective attention prioritizes the representational formats of relevant working memory contents and suppresses irrelevant contents remains an open question, however. Here, we addressed this issue using a multi-item working memory task involving a retro-cue. We recorded iEEG activity from patients with electrodes in VVS, PFC, and/or hippocampus. We report a dissociation of the representational signatures across working memory stages: During encoding, item-specific information was selectively observed in VVS and hippocampus, but not in the PFC. During maintenance, the geometry of activity patterns in the PFC was captured by the deepest layers of a recurrent DNN model. This effect was transient, locked to the presentation of the cue, and was specific to the beta (16-28Hz) frequency band. Our results suggest that PFC represents task-relevant information in an abstract representational format, dynamically affecting distributed representations in the VVS and hippocampus through top-down inhibitory mechanisms.

Talk 13: Mnemonic Signals in Visual Cortex During Working Memory Reflect Behavioral Precision, but Not Imagery

SimonWeber, Charité Universitätsmedizin Berlin

Visual Mental Imagery (VMI), the ability to generate depictive mental representations of visual content in the absence of external visual stimulation, has been proposed as a possible cognitive strategy to remember visual stimuli in Visual Working Memory (VWM) tasks. However, the ability to generate VMI varies substantially across individuals, leading to the hypothesis that individuals with high VMI vividness rely largely on sensory recruitment of visual areas to store visual inputs, while subjects with low VMI vividness might use alternative, non-visual cognitive strategies to solve VWM tasks. Accordingly, the cognitive strategy used to solve a VWM task would influence how much information about the stimulus is present in a given sensory store.

Here, we systematically investigated the influence of VMI vividness on the strength of VWM representations in the visual cortex, using fMRI. Two groups of participants with high and low VMI vividness scores were tasked to maintain the orientation of a stimulus grating over a delay period of 10 seconds. Orientations were continuously sampled across 0-180¡.

Using multitarget Support Vector Regression (mtSVR), a novel method to decode continuous periodic stimulus features from multivariate voxel patterns, we were able to reliably reconstruct remembered stimulus orientation from voxels in V1-V3 during the delay period. The reconstruction quality, however, did not differ between subjects with high and low VMI vividness. Instead, we observed a strong correlation between task performance and reconstruction quality. These findings suggest that behavioral precision, rather than VMI, predicts the strength of VWM representations in early visual areas.

Talk 14: Large-scale Estimation of Individualized Brain Models from M/EEG: Validity and Implications for Personalized tES

Matthew Singh, Washington University in St. Louis

A key goal of human brain mapping is to decipher how individual differences in brain signaling and dynamics relate to individual differences in cognition/behavior. Developing mechanistic models of individual human brains is one part of this endeavor. While the origins of fMRI functional-connectivity have been modeled extensively, individual-differences in transient, dynamical interactions (M/EEG) are less understood. Current whole-brain modeling approaches have proven valuable for hypothesis-generation but are not typically specified at the single-subject level or directly constructed from timeseries data (instead assuming parameters from DTI). Other approaches (e.g., DCM) have incorporate direct parameter-fitting for M/EEG but have been limited to a small number of brain regions due to limited optimization strategies. We aim to bridge this gap by presenting a new technique to directly estimate whole-brain dynamical systems models from M/EEG. Unlike structural connectivity, this procedure enables the estimation of directed connectivity between different cell-types (inhibitory vs. excitatory). The key to this success was the development of a new, well-validated optimization algorithm: the generalized backpropagation Kalman filter (gBKF). Using MEG data from the human connectome project, we constructed reliable and predictive models of individual human brains. Each model contains 100 brain-regions composed of excitatory and inhibitory populations with heterogeneous local connectivity/dynamics. Long-distance connections are described by directed excitatory-to-excitatory and excitatory-to-inhibitory connectomes. We demonstrate the validity of our approach with ground-truth simulation and cross-validation. We also tested the implications of these models for designing personalized brain stimulation under two objectives: changing alpha-band power or entraining activity to follow a reference signal.

Talk 15: Communication synchrony across corticostriatal connections: a temporal tier of circuit architecture

Cole Korponay, McLean Hospital/Harvard Medical School

At a given moment, millions of individual connections transmit communications from the frontal cortex to the striatum. This connection system subserves a host of complex human behaviors and is centrally implicated in many neuropsychiatric disorders. Prior work has intricately elucidated where communications from different areas of cortex arrive in the striatum and the relative strengths of these communications. Yet, the driving force of striatal neuronal activation and function is the receipt of many communications that arrive in synchrony, and currently there is little understanding of temporal coordination between communications across different corticostriatal connections. Here, expanding upon recent advances in 'temporally unwrapping' functional connectivity data, we reveal a set of tools to examine this 'temporal tier' of corticostriatal organization. First, we identify communities of cortex-striatum node-pairs whose moment-by-moment fluctuations in coordinated activity are preferentially synchronized with each other across time. We demonstrate that the striatum's 'communication synchrony' map parcellates into spatial communities that are orthogonal to those from classic maps of structural and functional connectivity, which may allow for functional integration in the temporal dimension. We also identify 'communication synchrony hubs' - striatal sites whose suite of communications with different areas of frontal cortex are disproportionately synchronized with one another. Furthermore, we find that a small number of spatially distant cortex-striatum node-pairs nevertheless maintain high communication synchrony, suggestive of a non-local corticostriatal circuit interaction that is largely unexplored. Finally, we find that interindividual differences in cortiostriatal temporal synchrony properties are related to variance in performance of a host of cognitive tasks.


Add to Calendar
04/23/2022 2:00 PM
04/23/2022 3:30 PM
Data Blitz Session 2
Data Blitz Session 2 will be held in person at the CNS 2022 Annual Meeting in San Francisco at the Hyatt Regency San Francisco Hotel. Located at 5 Embarcadero Ctr, San Francisco, CA 94111 in the Grand Ballroom B/C
Grand Ballroom B/C

CNS 2022 | Data Blitz Session 2

Talk 1: Induction of implicit emotional regulation affects directional connectivity of prefrontal and attentional areas

Miroslaw Wyczesany, Jagiellonian University

Implicit and automatized forms of emotional regulation, although understudied, are recently considered crucial for psychological well-being. In the experiment we implicitly induced self-regulation goal using a task with word scrambles and then asked participants to passively watch emotional stimuli. We hypothesized that the implicit modulation of affective responses acts already at the very early phase of stimuli processing. We also expected that the attentional networks contribute to these modulatory effects, while being controlled by bilateral dorsolateral prefrontal regions already shown to activate during both implicit and explicit emotional regulation.

The procedure has been previously validated using EEG and fMRI markers of emotional arousal. Current study involved 26 volunteers and consisted of neutral and goal-induction blocks, that differed with sentences to be put together from scrambles. They could be either neutral or include some self-control related content. EEG signal recorded during negative picture presentation was subjected to preprocessing, source reconstruction using minimum norm approach and directional connectivity analysis using the Directed Transfer Function estimated in the beta frequency band. Sources in bilateral medial frontal gyri along with nodes of the dorsal and ventral attentional networks were analyzed. Results demonstrated an increased bidirectional connectivity between both dorsolateral regions and also a higher top-down influence from these areas towards the right frontal eye field after induction of implicit control. It was accompanied by a strengthened effective connectivity within both attentional networks, which can reflect a modified attentional state that is responsible for modulating the depth of perceptual processing.

Talk 2: Emotional state dynamics impacts temporal memory

Jingyi Wang, University of California, Santa Barbara

Time unfolds continuously, but our temporal memories are fragmented. Emotion powerfully influences event memory, but precisely how it shapes temporal memory remains unclear. Prior studies show that contextual changes (event boundaries) typically lengthen the perceived temporal duration and reduce temporal order accuracy for events encoded across contexts. Emotional state changes differentially impact the neural correlates of emotional memory depending on the order in which they occur (Tambini et al. 2017). Here, we tested whether and how changes in emotional states modulate temporal memory across contexts. Participants viewed sequences of emotional and neutral images surrounded by a border. In an orthogonal design, we manipulated context (border color), emotional state (emotional picture valence), and the direction of emotional state shifts. After each picture sequence, participants judged the temporal interval and temporal order for image pairs presented within and across contexts. Overall, shifts in emotional states amplified subjective time, lengthening the perceived temporal interval between items presented across contexts. Critically, a shift from neutral to negative emotional states reduced temporal order accuracy for items presented across contexts. Conversely, temporal order memory was enhanced for items presented within negative (compared to neutral) emotional states. This study shows that emotional state dynamics matters when considering time-emotion interactions: First, emotional state shifts have a stronger effect on temporal interval memory compared to neutral (perceptual) contextual shifts. Moreover, while negative emotion can enhance temporal encoding overall, dynamic shifts to negative emotional states initially hinder temporal memory.

Talk 3: Common and selective representations of pain, emotion, and cognitive control in the insula

Mijin Kwon, Dartmouth College

The insula has been associated with numerous mental functions. However, direct evidence for functional selectivity and generalizability across insula subregions is limited. This study aims to address how insular sub-regions respond across four domains of mental processes: pain, positive emotion, negative emotion, and cognitive control.

First, we aggregated participant-level fMRI contrast maps from 36 studies and 540 participants. The sample was systematically constructed, with four domains, three subdomains per domain, three studies per subdomain, and 15 participants per study. This dataset was collected under the auspices of the Affective Neuroimaging Collaboratory, a collaborative data-sharing initiative.

We used Bayes factors to quantify evidence for positive activation versus null effects for each domain. The subsequent conjunction analysis identified insular voxels that are generalizable across domains or specific to a subset of domains. Regions selective for one domain included regions in the mid-posterior insula for pain, the mid-dorsal and ventral insula for positive emotion, the anterior-ventral insula for negative emotion, and the anterior/mid-dorsal insula for cognitive control. Respectively, 12.06%, 6.45%, 3.55% of insular voxels were selective for one, two, and three domains. We also found a subregion in the anterior insula engaged across all four domains.

Our results reveal a high degree of domain selectivity in the insula, but also support theories of integrative function in the anterior insula (e.g., Craig 2002). The combination of full voxel-level maps in a large sample and Bayesian evidence in support of null effects provides a more fine-grained picture of functional differentiation and integration in the insula.

Talk 4: Recognition of masked faces in the era of the pandemic: No improvement, despite extensive, natural exposure

Erez Freud, York University

Face masks became prevalent across the globe to minimize the effects of the COVID-19 pandemic. Previous research highlighted their negative impact on face recognition. An outstanding question is whether these effects would be attenuated following extended natural exposure to masked faces. This question is also relevant in the context of training effects on face recognition in natural settings. We utilized the Cambridge Face Memory Test (CFMT) in a cross-sectional study (total N=1,732), at six different time-points over a 20-month period (May 2020 - Jan 2022), alongside a 12-month longitudinal study (N=208). The results showed persistent deficits in recognizing masked faces across time-points. This was followed by a qualitative change reflected in a reduced inversion effect for masked faces. Additional experiments verified that the amount of individual experience with masked faces was not correlated with the mask effect. These findings provide compelling support for the notion that the face-processing system does not easily adapt to visual changes in face stimuli, even following prolonged, real-life exposure.

Talk 5: Self-representations across time become indistinguishable with distance from the present

Sasha Brietzke, Dartmouth College

A basic principle of perception is that as objects increase in distance from an observer, they also become logarithmically compressed in perception (i.e., not differentiated from one another), making them hard to distinguish. Could this basic principle apply to perhaps our most meaningful mental representation: our own sense of self? We conducted four studies that suggest selves are increasingly indiscriminable with temporal distance from the present, as well. In Studies 1-3, participants made trait ratings across various time points in the past and future. We found that participants compressed their past and future selves, relative to their present self. This effect was preferential to the self and could not be explained by the alternative possibility that individuals simply perceive arbitrary self-change irrespective of temporal distance. In Study 4, we tested for neural evidence of temporal self-compression by having participants complete trait ratings across time points while undergoing functional magnetic resonance imaging (fMRI). Representational similarity analysis (RSA) was used to determine if neural self-representations are compressed with temporal distance, as well. We found evidence of temporal self-compression in areas of the default network, including medial prefrontal cortex (MPFC) and posterior cingulate cortex (PCC). Specifically, neural pattern similarity between self-representations was logarithmically compressed with temporal distance. Taken together, these findings reveal a 'temporal self-compression' effect, with representations of temporal selves becoming increasingly indiscriminable with distance from the present.

Talk 6: Neuronal mechanism for the insight memory effect

Maxi Becker, Humboldt University

Increased evidence suggests that insight (sudden comprehension involving an AHA! experience) during problem solving enhances memory for those problems compared to those solved without insight (Danek et al., 2019; Kizilirmak et al., 2016). However, the underlying neuronal mechanism for this insight memory effect still remains unknown.

One study investigating visual insight found activity in the anterior medial temporal lobes (MTL) predicting later memory for those problems (Ludmer et al., 2011). However, because the solution was induced and the AHA! experience was not assessed it remains unknown whether MTL activity is related to insight.

We tested 30 subjects using a visual and verbal insight task in the MRI and asked them to rate how strongly they experienced an insight after each solution. Additionally, they performed a subsequent memory test (recognition of the insight problems from both tasks) four days later.

On a behavioral level, we replicated the insight memory effect: subjects were more likely to remember the insight tasks the stronger they rated their insight during solution in the scanner. On a neuronal level, we found overall activity in medial prefrontal cortex, bilateral anterior lateral temporal and MTL regions (including hippocampus) during solution predicting later memory for verbal and visual insight combined. Importantly, activity in bilateral hippocampus was parametrically modulated by the amount of experienced insight during solution.

This is first evidence for a neuronal mechanism of the insight memory effect: insight increases activity in the hippocampus necessary for binding content representations of the respective tasks into a durable memory representation.

Talk 7: Learning Goals Drives Retroactive Memory Enhancements for a Real-Life Experience.

Steven Martinez, Temple University

Motivation can flexibly enhance learning, such that memory for information directly associated with one's goals, as well as previously encountered conceptually-related information, is strengthened via consolidation (i.e., retroactive memory effects). These retroactive consolidation-related effects emerge for neutral events followed by emotional learning and for retroactive learning instructions (i.e., being told about a test after learning). Are similar retroactive memory effects seen when individuals are given learning goals to motivate future learning? That is, do instructions to learn about an upcoming event also enhance memory for similar prior events? 118 participants (ages 18-35: M = 20.7 years) completed a tour of a local haunted house that included four segments. The first two segments were each experienced naturally and without any instruction. During the next two segments, a control group experienced each segment naturally, while a goal-assigned group was given a learning goal prior to touring the segment. All participants completed immediate and one-week delay free recall assessments. Results showed a retroactive memory effect, such that those assigned a learning goal before touring the latter two segments exhibited enhanced recall for the prior two segments at a one-week delay, compared to participants who experienced all segments naturally (p = .01). Critically, there were no differences in memory during the immediate free recall assessment (p = .19), pointing to a specific role for consolidation. These findings suggest that engaging in goal-directed learning can retroactively enhance conceptually-related recall in naturalistic settings, and that this enhancement emerges exclusively after a period of consolidation.

Talk 8: The Distracted and Creative Brain: The Effect of Smartphone Notifications on Cognitive Control as a Function of Divergen

Joshua Upshaw, University of Arkansas

Prior work demonstrates that smartphone notifications can negatively affect cognitive control processes. Separately, people with greater creative thinking ability (i.e., divergent thinking, DT) demonstrate enhanced executive functioning. Few studies have assessed the effects of smartphone notifications on executive processes in creative people. Previous work found an association between smartphone use and increased creative expression. However, increased DT was linked to less frequent engagement in social media use. The present study examined the effect of smartphone notifications on cognitive control as a function of DT. DT was assessed with the Torrance Test of Creative Thinking. Cognitive control was assessed with a Navon letter oddball task. Here, a smartphone notification, control sound, or silence preceded each trial. In line with previous research, behavioral results indicated that people higher in fluency of DT had better cognitive control overall. Interestingly, higher fluency was also associated with slower responses on smartphone (vs. silent) trials. Cognitive control in terms of behavioral performance did not differ between the sound conditions as a function of DT. Higher originality of DT, on the other hand, was associated with a larger N2 oddball effect (i.e. better cognitive control) on smartphone trials. In conclusion, people higher in fluency of DT demonstrated better goal directed behavior yet were slower to respond when exposed to smartphone notifications. Higher originality of DT was linked with increased neural indices of cognitive control. Future work will examine how smartphone notifications affect different forms of creative cognition.

Talk 9: Brain Network Functional Connectivity Supporting Scientific Creative Thinking

Roger Beaty, Pennsylvania State University

Creative thinking has been linked to functional connectivity among the brain's default, salience, and control networks. However, past work has largely focused on 'domain-general' creative thinking (e.g., word associations), and comparatively less is known about the brain systems that support creativity in specific domains, such as scientific creative thinking. In the present fMRI study, participants (n = 47) completed a scientific hypothesis generation task (thinking of novel/plausible explanations for hypothetical scenarios) and a control task (thinking of synonyms to replace a word in a hypothetical scenario). We conducted a series of functional connectivity analyses to characterize the brain networks associated with hypothesis generation. Multivariate pattern analysis (MVPA) identified a whole-brain network associated with hypothesis generation, including hubs of the default (posterior cingulate cortex; PCC), salience (right anterior insula; AI), and semantic control (left inferior frontal gyrus; IFG) networks. Using these network hubs as seed regions, we found increased between-network functional connectivity during hypothesis generation (compared to the control condition), including stronger coupling between semantic control (IFG) and default regions (PCC and bilateral angular gyrus) and stronger coupling between salience (AI) and default regions, alongside weaker within-network functional connectivity. The present study represents the first network neuroscience-based investigation of scientific creative thinking. Our results indicate that, similar to general creative thinking, scientific creative thinking involves positive functional connectivity among the default, salience, and control networks, potentially reflecting a coordination of goal-directed and spontaneous/memory-based cognitive processes to construct original explanations for scientific phenomena. Our presentation will consider potential implications for science education.

Talk 10: Exploring Functional Brain Network Modularity in Educational Contexts

Adam Weinberger, Georgetown University

The brain's modular functional organization facilitates adaptability. Modularity has been linked with a wide range of cognitive abilities such as intelligence, memory, and learning. However, much of this work has (1) considered modularity while a participant is at rest rather than during tasks conditions and/or (2) relied primarily on lab-based cognitive assessments. Thus, the extent to which modularity can provide information about real-word behavior remains largely unknown. Here, we investigated whether functional modularity during resting-state and task-based functional magnetic resonance imaging (fMRI) was associated with academic learning and ability in a large sample of high school students. Additional questions concerned the extent to which modularity differs between rest and task conditions, and across spatial scales. Results indicated that whole-brain modularity during task conditions was significantly associated with academic learning. In contrast to prior work, no such associations were observed for resting-state modularity. We further showed that differences in modularity between task conditions and resting-state varied across spatial scales. Taken together, the present findings inform how functional brain network modularity ? during task conditions and while at rest ? can be used to predict and assess a range of cognitive abilities.

Talk 11: The integration of multiple social cues into expectations and their effect on perception in pain and vision

Rotem Botvinik-Nezer, Dartmouth College

Expectations about upcoming stimuli shape their perception. When integrating across multiple cues (e.g., ratings of other participants), the mean value across cues has been repeatedly shown to shape expectations and in turn perception. Some studies have found that this effect is weaker when the variance (uncertainty) across multiple cues is higher, in accordance with a Bayesian predictive-coding framework, but others have found that uncertainty increases pain perception irrespective of mean. Here, we hypothesized that these previous findings may result from over-weighting of extreme pain cues. We tested how humans integrate information into expectations to affect perception, and whether these processes are modality-specific or modality-general. We scanned 45 participants with fMRI while presenting 10 simultaneous cues, presented as ratings by other participants, followed by either a hot stimulus ('pain') or a flickering checkerboard contrast ('vision'). The mean, variance and skewness of the cues were experimentally manipulated. We found that the mean and skewness affected expectations in both modalities. Using Bayes Factors, we found that cortical and subcortical regions such as the basal ganglia, prefrontal cortex and parietal attention regions were related to the cue mean effect (high vs. low) in pain (substantial support for the alternative) and not in vision (substantial support for the null), but only low-level visual cortical regions were related to both modalities or only to vision and not to pain. Our results indicate that expectation-related effects are a combination of modality-specific and modality-general processes and that extreme values are overweighted when cues are integrated into expectations.

Talk 12: Cognitive Processing of a Common Stimulus Synchronizes Hearts, Brains and Eyes

JensMadsen, City College of New York

A wide range of neural, physiological and behavioral signals correlate between human subjects in a variety of settings. A great number of hypotheses have been proposed to explain this inter-subject correlation, however there is no clarity under which conditions or circumstances it arises, for which signals, or whether there is a common underlying mechanism. We hypothesized in this work that cognitive processing of a shared stimulus is the source of inter-subject correlation. We test this by presenting informative videos to participants in an attentive and distracted condition and subsequently measured recognition memory. Inter-subject correlation was observed for electro-encephalography, gaze position, pupil size and heart rate, but not respiration and head movements. The strength of correlation was co-modulated in the different signals, changed with attentional state, and predicted subsequent recognition of information presented in the videos. There was robust within-subject coupling between brain, heart and eyes, but not respiration or head movements. The theory that emerges is that inter-subject correlation is the result of cognitive processing of a shared stimulus, but can be observed only for those signals that exhibit a robust brain-body connection. While physiological and behavioral fluctuations may be driven by multiple features of the stimulus, correlation with other individuals is co-modulated by the level of attentional engagement with the stimulus.

Talk 13: Neural Decoding of Music Stimuli from Auditory Evoked EEG Responses

Adolfo G.Ramirez-Aristizabal, University of California, Merced

Recent studies have developed methods for decoding neural activity to interpret what participants are thinking. These studies demonstrated the ability of deep learning models to classify and reconstruct image classes from EEG recordings of participants viewing the images. The present study uses deep learning to classify music stimuli based on a dataset of EEG responses from participants passively listening to ten different songs (~ 4 mins each song). We used convolution neural networks (CNNs) to obtain 89% end-to-end classification performance for song labels from EEG inputs one second in length. Transfer learning using a standard computer vision model, ResNet-50, yielded 93% classification performance. To further understand the representations of music stimuli that can be learned from EEG recordings, we built a deep convolutional regressor to reconstruct the music stimuli spectra. In this model, the target was the corresponding time-aligned music stimuli spectrum for each corresponding one-second window of EEG recording. Quantitative validation of our music spectrum reconstruction involved using the outputs of the model as inputs to train a CNN classifier. The classifier was 81% accurate for test outputs trained using mel-spectrogram targets, and 72% accurate when using linear spectrogram targets. Finally, we inverted the reconstructed spectra to waveforms to be able to listen and see what features of the music stimuli were captured by the model, and which were not.

Talk 14: Reorienting of Attention Underlies the Intrusion of Unwanted Memories

Frederik Bergmann, University of Cambridge

While memory retrieval is often voluntary, memories can also intrude into awareness against our will. In fact, memory intrusions are a hallmark of many psychological disorders, such as anxiety, OCD, and PTSD. Even though unwanted memories are of major clinical significance and also constitute a ubiquitous phenomenon in daily life, they have proven challenging to study. Here, we isolated instances of involuntary retrieval in four fMRI studies of retrieval suppression using the Think/No-Think (TNT) paradigm. We show that when involuntary retrieval occurs, activity arises in the right dorsal intraparietal lobule (IPL), a region of the fronto-parietal control network, often active in working memory and attention tasks. Critically, we dissociated this response to involuntary retrieval from activation associated with recollection, which activated the left angular gyrus (AG). By conducting a Posner spatial-cuing task in the same participants, we demonstrate co-localization of intrusion related activity with areas engaged during reflexive reorienting of attention to invalidly cued items. Using multi-voxel pattern analysis, we demonstrated a significant similarity in the activation patterns associated with reorienting caused by visual stimuli and intruding memories. Taken together, these findings suggest that common neural mechanisms contribute to perceptual and mnemonic attention, and that these mechanisms are supported by the right dorsal IPL. We propose that the functional overlap originates from a similarity of spatial reorienting and mnemonic reorienting processes. As such an intruding memory or thought may capture attention like a goal-relevant visual stimulus does, signalling the demand for control processes to purge the intruded memory from awareness.

Talk 15: Shared representational geometry across the frontoparietal network supports credit assignment during social learning

Amrita Lamba, Brown University

Learning in naturalistic and ever-changing environments can unfold efficiently when the outcomes of our actions are attributed to the appropriate causes -a process known as credit assignment. Human social interactions, for instance, often entail learning precise causal mappings because the same action with different individuals can produce variable outcomes. In such situations, aptly assigning credit to our actions allows us to evaluate the boundary conditions of acquired learning rules and to build internal models of how associative links change across contexts. Little is currently understood about the cognitive or neural processes required for credit assignment. In the current study, participants performed a multi-player social learning task (the Trust Game) and a matched nonsocial task, which required learning how much money to invest with a set of partners/slot machines (targets) that varied in their reward rates. Results reveal that faster learning exhibited in the social task can be attributed to better credit assignment. We use a representational similarity analysis approach to compare the content of target-representations with behavioral performance. We find that the extent to which participants correctly execute target-selective responses predicts increasingly distinct neural representations in the frontoparietal network, aligning with prior work in macaques. Using trial-wise analyses we find that shared variance in the representational geometry during choice and value-encoding phases within this frontoparietal network was linked with better credit assignment. These findings provide novel evidence that the frontoparietal network is a critical hub for organizing how associative links should operate, allowing us to behave flexibly in naturalistic environments.


Add to Calendar
04/23/2022 2:00 PM
04/23/2022 3:30 PM
Data Blitz Session 3
Data Blitz Session 3 will be held in person at the CNS 2022 Annual Meeting in San Francisco at the Hyatt Regency San Francisco Hotel. Located at 5 Embarcadero Ctr, San Francisco, CA 94111 in the Bayview Room
Bayview Room

CNS 2022 | Data Blitz Session 3

Talk 1: Power Spectral Density Composition in Pre-Linguistic Infants

Beenish Mahmood, Rutgers University- Newark

Spontaneous brain oscillations have recently garnered much attention for their posited utility as early biomarkers of atypical neural development. Previous studies have shown that spontaneous electroencephalographic (sEEG) data from infants that are at high risk for developmental delay show differences in spectral power composition when compared to infants who are at low or no risk, and that these differences may correspond to later cognitive deficits. However, no study to date has examined whether spectral power composition can be altered as a result of early behavioral intervention. In the present study, we examine sEEG data collected from infants at 7, 9, 12 and 18 months. Infants in the Active group were presented with an interactive acoustic experience (IAE) between 4-7 months of age, while age-matched infants in the Na•ve group were not. We hypothesized Power Spectral Density (PSD) values in a given frequency range would differ by group. Absolute PSD values were calculated for each infant for alpha, beta, theta, and gamma frequency ranges and groups were compared across ages. Consistent with previous literature, preliminary results showed that Frontal PSD values in the higher frequency ranges increased with age. Group differences were seen in the higher frequencies at 12 and 18 months of age. Additionally, group differences were seen in cognitive assessment scores at 18 months of age. These findings begin to quantify how early experience can support the development of networks in a manner that can enhance later cognitive and language ability.

Talk 2: Functional selectivity and structural connectivity of the cortical language network are intact in dyslexia

Jayden Lee, Boston University

Developmental dyslexia is characterized by a specific reading impairment for which the underlying cause is still heavily debated. While reading-related neural circuitry has been extensively studied in neuroimaging studies of dyslexia, the function of core language comprehension regions independent from the task of reading remains largely unexplored. In this study, we employed fMRI and diffusion tractography analyses to characterize the language processing network in dyslexia. First, we used the spoken language to identify the cortical regions selectively responsive to higher-level linguistic processing in adults with dyslexia and typically reading controls. Participants listened to intact naturalistic speech and acoustically-degraded speech (Scott et al., 2017). We compared the activation patterns between the two groups using both univariate and group-constrained subject-specific analyses and found (1) no significant differences between groups for whole-brain activation to the intact vs. degraded speech contrast, (2) similar parcellation of the language network, identifying the same functional nodes in both groups, and (3) no group difference in the selectivity of the functional regions-of-interest (fROIs) for language vs. spatial working memory tasks. This pattern of results suggests that the functionality of the core language network is fundamentally similar in dyslexia. Finally, we used probabilistic diffusion tractography to examine structural connectivity among the individually-defined language fROIs. The resulting ROIxROI structural connectivity matrix showed no differences between groups. These findings suggest that the functional neuroanatomy of the brain's ecological spoken language comprehension network is fundamentally intact in dyslexia.

Talk 3: Periodic Chunking of Language: Evidence from Self-Paced Reading

Lars Meyer, Max Planck Institute for Human Cognitive and Brain Sciences

Introduction: Humans segment linguistic input into cognitive units to achieve comprehension. It has been argued that the period of the underlying rhythmic neuronal processes affects the pace of segmentation (Meyer et al. 2020; Henke and Meyer 2021). Periodic brain activity has been observed to accompany phrase-level language processing (Ding et al. 2016). It remains unclear whether this is behaviorally relevant for segmentation; moreover, it is not clear what type of cognitive units it reflects. We test whether segmentation behavior also shows a periodic pattern; we also assess whether periodicity links to cognitive units as formalized through natural language processing. Experiment: We analyzed self-paced reading (SPR) data from 180 participants (Futrell et al. 2021). To assess periodicity, we performed frequency analysis on word-by-word reaction times (RT). To focus on segmentation, we performed differencing of the time series on the RTs, highlighting abrupt changes from slow to fast RTs; these have been related to segmentation previously (Tosatto et al. 2021). We then statistically predicted the slow - fast changes from the cognitive units formalized by the chunk-and-pass model (Anderson et al. 2019). Results: Frequency analysis showed that abrupt slow - fast changes occur with a period of ~1 Hertz. Moreover, they cluster at sentence boundaries (Just and Carpenter 1980; Rayner et al. 2000), but also at the boundaries of the cognitive units predicted by the chunk-and-pass model. Our study shows that segmentation of linguistic input into second-long cognitive units is a periodic behavior, possibly related to the periodicity of the underlying neuronal processes.

Talk 4: Listeners extract spectral and temporal information from the mouth during naturalistic audiovisual speech

Cody Zhewei Cao, University of Michigan

Seeing a speaker's face helps speech perception during challenging listening conditions. But, how do we recover speech information from the face? Eyetracking research shows that while listeners' fixations are distributed throughout the face during clear speech perception, gaze fixation to the mouth increases when speech is noisy, implying that visual signals from this region are necessary to restore auditory speech information. Indeed, it has previously shown that listeners can recover spectral information using speakers' mouth width and that lip-closure helps parse the boundary between words. However, it remains possible that these two statistical features could be extracted from other regions of the face, or whether viewing the mouth is sufficient for restoring spectral and temporal speech information. First, using eyetracking, we examined where listeners fixate during speech that has been spectrally or temporally filtered. Across both conditions, subjects successfully recovered the degraded auditory signals by uniformly fixating on the mouth. Second, we occluded the upper or lower face to test whether audiovisual recovery occurs in the absence of the mouth. Preliminary results suggest that listeners extract both temporal or spectral information from visual mouth cues. Finally, using parameters of the mouth extracted from deep-learning features of the face, we are testing whether spectral and temporal speech features are extracted from different mouth movements. We hypothesize that horizontal movements are more important for spectral recovery and vertical movements more so for temporal recovery. Taken altogether, we show the mouth is necessary and sufficient for the recovery of temporal and spectral information.

Talk 5: Naturalistic viewing conditions can increase task engagement and aesthetic preference but have only minimal impact on EE

Dominik Welke, Max-Planck-Institute for Empirical Aesthetics

Free gaze and moving images are typically avoided in EEG experiments due to the expected generation of artifacts and noise. Yet for a growing number of research questions, loosening these rigorous restrictions would be beneficial. Among these is research on visual aesthetic experiences, which often involve open-ended exploration of highly variable stimuli.

Talk 6: Interleaved vs Blocked Training Format Effects on Optimizing Music-Related Transfer to Auditory Skills

Ozgen Demirkaplan, Memorial University of Newfoundland

Short-term musical training, over the course of a few weeks or months, has been shown to enhance both auditory and cognitive abilities. These findings suggest that music training may be useful as a form of cognitive or auditory rehabilitation. Training programs across a variety of fields are based on maximizing performance through practice, often by repetition. An extensive body of research has shown, however, that gains made during practice do not necessarily persist. If music training is to be used as a form of rehabilitation, it is critical to optimize the training so that it is effective in the shortest time possible. Accordingly, the goal of this research was to compare two formats of music training on their capacity for transfer to auditory abilities. The Blocked Group practiced playing a novel melody on a piano repetitively before moving on to another novel melody. The Interleaved Group practiced the same melodies for the same time, but the practicing order of each melody was randomized within each practice session. Consistent with the literature, the Interleaved group, compared to the Blocked Group, exhibited lower performance as the practice sessions went on for two weeks. However, the Interleaved group, ERP results in an auditory oddball paradigm show improvement in early neural responses for small pitch detection. Despite this difference, the benefit to auditory and cognitive abilities was comparable in both groups, suggesting that the training format for such short term may not impact the potential rehabilitative benefits of music training.

Talk 7: Effector-independent prehension action representations

Naama Zur, Georgetown University

Performance of motor actions, specifically - reaching and grasping, is well characterized for the hand. However, performance of these motor actions by other limbs (i.e. foot) can be used to look into whether there are shared regions activated during performance of these actions in different limbs. we used fMRI to study the abstraction of action representations as participants performed visually-guided reaching and grasping towards a cuboid with either the right hand or right foot. Cuboid grasping additionally varied in size (large and small aspect of the object). We found that there was a preference for action type (grasping vs. reaching) independent of the effector, the body part performing the action, in the anterior intraparietal sulcus (aIPS). The same preference was found for foot grasping in two individuals born without hands, who habitually use their foot for manual actions. Further, preliminary findings show that multivariate decoding of grasping size across body parts (training the classifier with the hand trials and testing with those of the foot, and vice versa) was possible along the IPS. These results support previous findings showing effector-independent action representation across the two hands, and extend them to a remote body part the foot. Together, they suggest that aIPS is involved in the performance of grasping independently of the performing limb. These findings further support the notion that motor actions are represented at an abstract level independent of sensorimotor parameters.

Talk 8: The Adapted Body: visual adaptation aftereffects in the perception of body parts

Francesca Frisco, University of Milano-Bicocca

Adaptation aftereffects refer to perceptual changes due to exposure to a previous stimulus. An established size-contrast aftereffect is the Uznadze illusion: after simultaneous, prolonged exposure to two objects different in size, an object of the same size is perceived as smaller after a larger adaptor and larger after a smaller adaptor. However, it is still debated whether only low-level sensory or also high-level cognitive processes mediate the aftereffect. We investigated whether the Uznadze visual aftereffect is the same for visual stimuli representing body parts (which activate a dedicated neural representation in the brain) and for non-bodily shapes. In two experiments, pairs of bodily (hands=BS) or non-bodily (shapes=N-BS) stimuli were presented to each participant in separate sessions. The procedure consisted of an initial adaptation phase in which a large and a small stimulus were simultaneously presented 15 times on both sides of the fixation. Then, in the test phase, participants judged which stimulus was bigger between a standard stimulus with a constant size and a test stimulus of variable size. In Experiment 1, we found a stronger aftereffect for BS than for N-BS, but only in participants who first performed the N-BS session. In Experiment 2, where the similarity between BS and N-BS was reduced, we found a stronger aftereffect for BS than for N-BS, regardless of session order. In conclusion, high-level cognitive processes related to the stimulus meaning could mediate the perceptual aftereffect by increasing it for BS or when, after a pre-exposure to BS, non-BS are interpreted as BS.

Talk 9: Dreaming in the dark: Exploring neural correlates of dreams lacking visual imagery

Karen Konkoly, Northwestern University

Visual imagery is extremely prevalent in REM-sleep dreams. Yet, how and why the sleeping brain generates visuals during sleep remains poorly understood. Here, we sought to investigate this topic by studying behavior and brain activity during dreams without visual content. Closing one's eyes while awake drastically reduces visual input and produces a characteristic EEG change, an increase in posterior alpha power. During REM sleep, alpha power is typically lower than during wake, and REM eye movements have been proposed to aid in the creation of dream imagery. Do any features of REM sleep change when dreams lack visual imagery? To investigate this question, we studied individuals aware of dreaming while still asleep. These lucid dreamers attempted to close their eyes within their dreams and report on their visual experiences in real time, drawing on the methodology of two-way communication with dreamers (Konkoly et al., 2021). They produced behavioral responses by sniffing, measured via a nasal cannula. To date, 7 participants completed 16 overnight sleep sessions, and in 11 sessions, participants reported attempting to close their eyes in a lucid dream. Concurrent EEG alpha did not increase reliably (as it does during wake), and further EEG analyses are in progress. By exploring these unique aspects of dreaming, we hope to gain new information on the neural underpinnings of vision and visual inhibition in REM sleep versus wake, thereby shedding light on the general question of how dream visuals are generated.

Talk 10: Perceptual expectations and false percepts activate distinct layers of the visual cortex

Joost Haarsma, University Bordeaux

Hallucinations are a key-feature of various psychiatric and neurological disorders, yet the neural mechanisms remain poorly understood. There are a wide range of theories that attempt to explain hallucinations, with some emphasising the role of top-down prediction signals, while others emphasise the role of spontaneous bottom-up neural activity. In the present study we aim to test these theories by measuring stimulus specific-activity in the input and feedback layers of the early visual cortex using laminar fMRI while perceptual predictions are generated and false percepts are reported. 25 individuals performed a difficult perceptual discrimination task under noisy conditions while a pre-learned cue predicted which grating orientation would appear. Crucially, on 50% of trials the to-be-detected visual stimulus was omitted, and only visual noise was presented, causing participants to rapport false percepts. We found that perceptual expectations and high confidence false percepts evoked orthogonal layer-specific patterns of activity in the early visual cortex. That is, perceptual expectations activated the deep layers of V2 even in the absence of an actual stimulus, although they did not induce false percepts. In contrast high confidence false percepts were reflected in the middle layers of V2, suggesting these may have resulted from local fluctuations in bottom-up neural activity. We further found that these high confidence false percepts were positively correlated with the presence of every day hallucinations in a larger online study. In conclusion, we demonstrate that false percepts can arise from input layer activity nuancing the importance of these feedback signals in driving false percepts.

Talk 11: Towards a Mechanistic Account of Perceptual Curiosity

Michael Cohanpour, Columbia University

(*indicates co-PIs) Realistic environments are rife with uncertainty. Often the uncertainty is sensory, requiring us to clarify ambiguous stimuli. Is the shadowy figure ahead a bear or a rock? Perceptual curiosity, the intrinsic desire to reduce sensory uncertainty, may modulate how we navigate such situations. Here, we asked how perceptual curiosity is related to neural representations of confidence and sensory uncertainty. During fMRI, participants performed a novel behavioral task in which they viewed a distorted image ('texform') of an animal or inanimate object and rated their confidence in recognizing the image and their curiosity to see the original (undistorted) image. We show that, like curiosity about trivia questions, perceptual curiosity has a negative quadratic (inverted-U) relation with confidence. Moreover, perceptual curiosity was predicted by two separate neural signatures of sensory uncertainty. One was a multivariate representation of uncertainty about stimulus category in occipitotemporal cortex (OTC). The other was a univariate response in subregions of the prefrontal cortex, which correlated with subjective reports of confidence and predicted curiosity independently of the OTC. The results reveal the neural network mechanisms by which uncertainty predicts subjective judgments of perceptual curiosity.

Talk 12: Ventral visual cortex signal volatility correlates to sight restoration outcome

Nanak Nihal Khalsa, Georgetown University

Patients who grew up without sight have varying visual outcomes on dimensions such as acuity and contrast sensitivity after sight-restoration surgery later in life. There are no known causal explanations for this variability, and not even any correlates that could be used to predict post-operative outcomes. Attributes such as age at treatment have proven surprisingly ineffective as predictors. As part of Project Prakash, we have collected functional outcome measures as well as neuroimaging data for several children who experienced late sight onset. Here we report our investigation of covariations between these two data-sets. Specifically, we examined the role of variability of ongoing signal fluctuations within the visual system. We measured mean squared successive difference (MSSD), a measure of temporal variability robust to slow fluctuations, in fMRI in Project Prakash patients who underwent surgery for congenital dense bilateral cataracts. Temporal variability in early ventral visual cortex was found to be anticorrelated with visual outcome quality, i.e. larger fluctuations in these regions were associated with lower visual acuity gains following cataract removal. This suggests that when beginning to receive visual input long after early developmental periods are over, ventral visual cortex shape processing may be hindered by difficulties in reliably responding to incoming signals, resulting in low neural SNR.

Talk 13: Selective and Non-Overlapping Intracranial Electrophysiological Responses to Bodies and Faces in the Human Fusiform Gyru

Claire Perry, Stanford University School of Medicine

The human fusiform face area (FFA) and the closely related fusiform body area (FBA) can be localized using non-invasive imaging methods. However, controversy exists as to the relative anatomical locations and functional distinctions between the two areas in a given individual brain. To address this issue, we leveraged the high anatomical resolution of electrocorticography (ECoG) recordings with grids of electrodes over the fusiform gyrus. We collected data in eight patients who were shown images of six categories including human bodies, human faces, nonhuman bodies, and nonhuman faces. From these recordings, we were able to identify distinct and selective electrodes in the FFA and FBA of each subject which showed minimal responsiveness to other stimuli. Most importantly, the relative location of the FFA and FBA varied in each subject's brain, a finding that may account for the ambiguity and controversy in previous research using group level analyses. Interestingly, FFA electrodes showed an earlier and greater response to human faces than did the FBA electrodes to human bodies on the individual subject level (p < 0.01 with an unpaired permutation test). Moreover, consistent with previously reported findings, nonhuman faces generated a later and diminished response when compared to human faces in the FFA and little to no response in the FBA, as was true for nonhuman body responses in the FBA. Taken together, these findings all suggest that the FFA and FBA show unique functional and anatomical profiles with minimal overlap at the individual brain level.

Talk 14: A new tripartite landmark in posterior cingulate cortex

Ethan Willbrand, University of California, Berkeley

Understanding brain structure-function relationships, and their development and evolution, is central to neuroscience research. Here, we show that morphological differences in posterior cingulate cortex (PCC), a hub of functional brain networks, predict individual differences in macroanatomical, microstructural, and functional features of PCC. Manually labeling 4,319 sulci in 552 hemispheres, we discovered a consistently localized shallow cortical indentation (termed the inframarginal sulcus; ifrms) within PCC that is absent from neuroanatomical atlases. Subsequent analyses revealed four main findings. First, the ifrms identifies a focal cluster of high cortical thickness and light myelination in PCC. Second, the ifrms co-localizes with, and predicts the location of, a functional sub-region of the cognitive control network (CCN). Third, the ifrms is identifiable in juvenile and healthy older adults, as well as in 48% of chimpanzee hemispheres examined. Fourth, morphological analyses showed that unique properties of the ifrms differ across the lifespan and between hominoid species. Altogether, our results reveal that the ifrms is a new tripartite landmark in human PCC that is developmentally and evolutionarily relevant. These findings help reconcile prior discrepancies regarding where the default mode network ends and the CCN begins in PCC. Intriguingly, the consistency of the ifrms also debunks the uniqueness of the morphology of Einstein's PCC. These findings support a classic theory that shallow, tertiary sulci serve as landmarks in association cortices, laying the groundwork for future investigations of PCC functional neuroanatomy. They also beg the question: how many other cortical indentations have we missed?

Talk 15: Modelling representations of continuously shifting stimuli: a novel sliding window approach

Brandon Forys, University of British Columbia

In everyday life, we are exposed to ever-shifting environments, from the movement of trees in the wind to the slowly changing colours of a sunset. Our experience and perception of these changes is continuous, and the brain representations of these states transition seamlessly. Yet, current methods of analyzing these neural representational patterns (e.g., representational similarity analysis; RSA) divide these experiences into artificially discrete time points, as the general linear model (GLM) on which they are founded is limited in its temporal capacity to model sequential changes to the hemodynamic response. In order to capture representations of continuously shifting stimuli, we propose a novel method we are currently working on for deploying RSA to assess gradual changes in neural representations of visual stimuli. Specifically, we propose the inclusion of blood oxygenation level-dependent beta weights derived from multiple GLMs, each modelling the same visual experiences at fixed but offset time windows. For example, five GLMs, each modelling a series of 2 s experiences with an absolute offset from each other of 400 ms can allow for 50 unique response betas in a 20 s event. RSA conducted on the subsequent betas can identify the rate of change in the representational pattern across time in response to gradually changing experience. In pairing these results with behavioral results, we show how this method can address outstanding questions about the rate and nature of representational shifts during visual recognition tasks.


Add to Calendar
04/23/2022 2:00 PM
04/23/2022 3:30 PM
Data Blitz Session 4
Data Blitz Session 4 will be held in person at the CNS 2022 Annual Meeting in San Francisco at the Hyatt Regency San Francisco Hotel. Located at 5 Embarcadero Ctr, San Francisco, CA 94111 in the Seacliff Room
Seacliff Room

CNS 2022 | Data Blitz Session 4

Talk 1: Role of the Ventromedial Prefrontal Cortex in Mnemonic Discrimination

Claire Lauzon, York University

Pattern separation-the neurobiological process of making overlapping mnemonic information more distinct-has been shown to depend on the hippocampus. The Mnemonic Similarity Task (MST; Stark et al., 2015) is a behavioural estimation of this process wherein participants must distinguish previously learned images of everyday objects from novel, highly similar images (lures) and dissimilar images (foils). Several studies have validated the use of the MST to infer hippocampal integrity, but the extent to which this task relies on other brain regions is unclear. The ventromedial prefrontal cortex (vmPFC) is involved in strategic aspects of memory encoding and retrieval, including 'feelings of rightness' and contextual memory organization. Individuals with lesions to the vmPFC have previously demonstrated inflated confidence in erroneous memories, possibly due to faulty meta-mnemonic monitoring after injury to this region. Using the MST, we examined if behavioural pattern separation critically depends on the vmPFC by testing individuals with selective lesions to this region. We also examined whether confidence in the accuracy of individual responses varied depending on stimulus type. Individuals with vmPFC lesions were selectively impaired in discriminating studied items from similar lures but not novel foils, resembling findings in individuals with hippocampal compromise. Furthermore, individuals with vmPFC lesion were highly confident when mischaracterizing similar lures, but not when responding incorrectly to novel foils and studied images. As such, deficits in lure discrimination may reflect faulty meta-mnemonic monitoring and an elevated 'feeling of rightness'-functions influenced by the vmPFC. These findings provide novel insight into non-hippocampal contributions to mnemonic discrimination.

Talk 2: Comparison of hippocampal activity during picture recognition and naming helps disentangle the language-memory crossroad

Joey Hsu, University of Pittsburgh School of Medicine

The hippocampus has classically been regarded as subserving mnemonic processes such as recognition memory. However, focal epilepsy patients who have undergone anterior temporal lobectomy often experience postoperative word retrieval deficits. In addition, neurophysiological work has demonstrated hippocampal activation during picture naming tasks, fueling a discussion on a role of the hippocampus in an expanded language network. To explore this potential intersection of memory and language processes, we compared intracerebral hippocampal recordings from 20 focal epilepsy patients while they performed visual picture recognition and naming tasks during presurgical evaluation. The recognition memory task elicited a negative-positive complex occurring around 400 and 600 ms, which was modulated as a function of repetition; repeated stimuli (retrieval) elicited larger P600 than their first presentation (encoding) or lures. This 'old/new' effect was observed in half of the patients. Picture recognition and naming activities were then compared. In the anterior hippocampus, a significantly greater proportion of patients demonstrated ERP differences between retrieval vs. second naming block than between encoding vs. first naming block (75% vs. 35%, Fisher's exact test: p < 0.05; N = 20). In the posterior hippocampus, no significant difference in proportions was observed (N = 12). Finally, in the entorhinal cortex, a trend-level difference was observed (67% vs. 25%, p < 0.1; N = 12). These results suggest that hippocampal involvement in naming might be incidental during first presentation and a result of different retrieval processes during second presentation. Furthermore, these differential processes are mainly linked to the head of hippocampus.

Talk 3: The contribution of anterior and posterior hippocampal connections to individual differences in memory specificity and g

Lea Frank, University of Oregon

The hippocampus aids in remembering individual experiences (memory specificity) as well as linking across those experiences to form generalized knowledge (memory generalization). How can the hippocampus support both memory functions? Prior work indicates functional dissociations between the anterior and posterior hippocampus such as distinct connectivity profiles and distinct levels of granularity in memory representations. In this study, we tested the hypothesis that anterior and posterior hippocampal connections will differentially predict behavior, such that anterior hippocampal connections will track with memory generalization while posterior hippocampal connections will track with memory specificity. A large sample of participants (n=167) completed a battery of memory assessments across two sessions that included measures of both memory specificity and generalization. Factor analysis was used to derive individual scores of memory specificity and generalization abilities. A subset of participants (n=62) returned for a third session to undergo MRI, including a resting-state scan. Individual differences in whole-brain connectivity with the anterior and posterior hippocampus during rest were correlated with individual differences in memory abilities. We found that memory generalization ability was predicted by only anterior hippocampal connectivity. Unexpectedly, no hippocampal connections predicted overall memory specificity ability. Instead, different hippocampal connections appeared to predict performance on individual tasks. These findings provide new probes into the hypothesized functional dissociations of anterior and posterior hippocampus and different aspects of memory.

Talk 4: Disruption of anterior temporal lobe reduces distortions in memory from category knowledge

Alexa Tompary, University of Pennsylvania

Memory retrieval does not provide a veridical recapitulation of past events, but instead an imperfect recombination of event-specific details and general knowledge.  Integrating these sources of information introduces systematic errors if they are discrepant. However, it remains unclear whether retrieval relies on a mixture of signals from different memory systems, one supporting event-specific details and one that supports more general knowledge, like prior semantic knowledge. We developed a protocol that enabled us to tease apart the disparate reliance on event-specific details and semantic knowledge when retrieving a single event (Tompary and Thompson-Schill 2021). In these experiments, participants could integrate across newly encoded image-location associations to learn that images' locations often clustered by their category membership. We found that when an image's encoded location was far from its cluster of category neighbors, participants placed it closer to its category cluster at retrieval, suggesting a bias in memory due to semantic knowledge. Here, we use this procedure with transcranial magnetic stimulation (TMS) to the left anterior temporal lobe (ATL). We predicted that disrupting ATL before retrieval would attenuate the extent of bias. With this within-subjects design (N=36), we found that TMS to ATL resulted in less bias in location memory, but only for atypical category members. Critically, the magnitude of error in memory was not impacted, suggesting that the fidelity of a memory can be decoupled from its distortion by semantic knowledge. This provides novel evidence that episodic memory retrieval is supported by the joint influence of separable, independent memory systems.

Talk 5: Do entrained low-frequency oscillations support memory formation through coupled gamma activity?

Paige Hickey, Tufts University

The brain spontaneously tracks rhythmic temporal regularities in the environment, such as music, through synchronization of low-frequency neural oscillations to the beat. In a recent study, we found that evidence that this low-frequency neural entrainment is associated with rhythmic modulations of memory encoding (Hickey et al., 2020). Specifically, we found that individuals with stronger entrainment of a background beat during encoding demonstrated better subsequent memory for visual images presented in-synchrony versus out-of-synchrony with the beat. Additionally, within subjects, low-frequency entrainment to the musical rhythm was greater for subsequently remembered compared to subsequently forgotten images. However, an outstanding question is the mechanism by which these entrained low-frequency oscillations support encoding. We hypothesized that entrained low-frequency oscillations influence memory encoding by modulating the amplitude of higher-frequency gamma oscillations through cross-frequency phase-amplitude coupling (PAC). To test this hypothesis, we analyzed data from our prior EEG study in which participants incidentally encoded on-beat or off-beat images in the presence of a background musical rhythm. Results indicated that there was significant PAC between the phase of the entrained low-frequency oscillation and amplitude of high-frequency gamma oscillations. In addition, gamma power was greater at the time of stimulus presentation for on-beat compared with off-beat trials. These findings provide preliminary evidence that entrained oscillations influence memory through modulating high-frequency gamma activity.

Talk 6: Identifying the Neural Mechanisms of Zone State Performance using Time-varying Functional Connectivity Methods.

Dolly Seeburger, Georgia Institute of Technology

There is ambiguity in the literature about how large-scale brain networks contribute to focused attention. Part of the problem comes from a common assumption that attention is like a light switch either on or off, but a better metaphor is to equate attention to a candle that flickers even when it's lit, the flame varies. With this metaphor, we can better understand fluctuations in attention over time. This continuous change in attention is consistent with the dynamic changes in functional connectivity between brain regions involved in internal and external allocation of attention.

Talk 7: Entorhinal grid-like signals reflect temporal context for human timing behavior

Ignacio Polti, Kavli Institute for Systems Neuroscience, NTNU

The entorhinal cortex (EC) supports the encoding of task regularities. A critical function may be the encoding of temporal context (i.e., forming integrated relational representations of co-occurring events and stimuli). A key neural component in the EC are grid cells, whose activity likely exhibits a six-fold rotational symmetry as a function of gaze direction as measured by functional magnetic resonance imaging (fMRI). Here, we combined fMRI and a time-to-contact estimation task to test whether temporal context modulates this grid-like fMRI activity in the human EC. In addition, we characterized in detail the relationship between trial-wise entorhinal activity and participants' task performance. We found that activity in the EC reflected biases in timing behavior, and that the cross-validated amplitude of grid-like signals indeed depended on the timing errors consistent with temporal-context encoding. These findings suggest that the human EC contributes to adapting internal timing mechanisms to the temporal statistics of the environment in accordance with the predictions from Bayesian models of time perception.

Talk 8: Temporal context drift produces spacing effects in an entorhinal-hippocampalmodel via error-driven learning

James Antony, University of California, Davis

The entorhinal cortex (EC) may send information about temporal context to the hippocampus (HC) during the formation of episodic memories. Specifically, EC neurons 'drift', or change activity slowly across multiple timescales, from seconds to hours. We argue that modeling this drift could help explain the the spacing effect, or how distributing learning over time improves long-term memory. Specifically, differences between stored and current temporal representations should produce greater error and subsequently greater error-driven learning, strengthening elements common across learning examples, like slower-drifting temporal representations and/or paired associates; such a process would result in either temporal abstraction (strengthening of more abstract timescales) or decontextualization (direct cue-target strengthening). Here, we advanced a neurobiologically realistic model of the EC and HC with time-drifting properties that can explain spacing effects. We trained the model to learn cue-target pairs alongside temporal context vectors, simulating drift in these vectors over multiple timescales. We tested the model on paired associates after varying drift between successive learning episodes and/or before final retention intervals. Greater drift reproduced numerous spacing effects, including better memory benefits at longer retention intervals. Dissecting model mechanisms revealed that greater drift increased error (more error-driven learning), stronger weights in slower-drifting temporal context neurons (temporal abstraction), and greater strengthening of direct cue-target representations (decontextualization) in HC. Intriguingly, these results suggest that decontextualization - generally ascribed only to the neocortex - can occur within the hippocampus itself. Altogether, these results provide a mechanistic formalization for established learning concepts such spacing effects and errors during learning.

Talk 9: The Effect of Deterministic Reward on Voluntary Task Switching as revealed by Eye Tracking and EEG

Juan Balcazar, Texas A&M University

Cognitive flexibility, defined as the brain's ability to adapt to an environment, is frequently measured via the voluntary task-switching (VTS) paradigm. In VTS, a higher proportion of switch trials compared to total is said to reflect higher cognitive flexibility. We examined VTS under deterministic reward conditions, in which participants used feedback to learn fixed point contingency values (1 or 10 cents) associated with cues. We used eye-tracking to investigate the pupillary dynamics and gaze fixations across each trial. In addition, EEG was collected and will be analyzed. Our preliminary results show pupil dilation was significantly greater for trials where reward increased vs. decreased relative to previous trial and when reward increased compared to when reward remained low, which is consistent with previous studies. While we had predicted increased fixations during the cue-to-target interval to the target location following high vs. low reward cues, preliminary analyses showed no differences. Increases in reward relative to the previous trial were associated with increased pupil dilation, suggesting increased anticipation when reward increased. For our ERP analyses, we hypothesize an enhanced P2 for high vs. low reward trials and a continuous strengthening of the CNV over time. These findings help reveal the underlying cognitive and neural mechanisms associated with VTS as a measure of overall cognitive flexibility.

Talk 10: Altered Associations between Motivated Performance, Striatal Activation, and Frontostriatal Connectivity during Reward A

Jason Smucny, University of California Davis

The contributions of disrupted brain activity to altered behavior during motivated performance in schizophrenia (SZ) are poorly understood. Using the drift diffusion model (DDM), we examined relationships between the drift rate (DR) parameter as an index of evidence accumulation with brain function during the Incentivized Control Engagement Task (ICE-T). The ICE-T is a delayed match-to-sample task that dissociates reward motivation and top-down cognitive control by incorporating two difficulty conditions (difficult and easy) and two reward conditions (reward vs. neutral). 3T fMRI ICE-T data during reward anticipation were analyzed from 48 healthy controls (HCs) and 52 people with recent onset schizophrenia (SZ). Individual performance was fit using a DDM, allowing the DR (a parameter that depends on both accuracy and reaction time) to vary between task conditions. Associations between dorsolateral prefrontal cortex (DLPFC), inferior parietal, and dorsal striatal activation with DR as well as between frontostriatal connectivity with DR were examined using mixed effects models. DRs were significantly higher for the reward (vs. neutral) condition and significantly lower in SZ (vs. HCs). Greater positive relationships were observed between activation and DR during reward (across difficulty) in HC vs. SZ as well as between frontostriatal connectivity and DR during reward-difficult in HC vs. SZ. Loss of frontostriatal connectivity during the reward-difficult condition predicted negative symptoms in SZ. These findings suggest that motivated behavior and frontostriatal circuit function are less well-linked in SZ under specific task conditions and have implications for understanding the neurobiology of reward anticipation deficits in the illness.

Talk 11: Mechanisms of rewarded categorical memory organization from neural network modelling.

Elizabeth Horwath, Temple University

A large body of research illustrates the prioritization of goal-relevant information in memory (Adcock & Murty, 2017); yet investigations into the mechanisms underlying the organization of valuable memories are sparse. We have recently shown that rewarded memories are not organized by temporal features of learning, but instead by higher-order value information (Horwath et al., in-prep). However, the mechanisms that support adaptive memory organization are unknown. Using this behavioral data, we simulated the enhancement of memory and clustering of recall by reward using a reward-modulated variant of the Context Maintenance and Retrieval Model (CMR; Polyn et al., 2009). Model simulations showed that a combination of modulating learning rate and source features of CMR most accurately depict the memory accuracy and reward clustering findings. The dynamic learning rate applied for rewarded items suggests that more strongly associating important information with its context leads to greater likelihood of recall and clustering of rewarded information. The source layer determines which reward context (i.e., high or low) each item belongs to. This parameter suggests that creating a categorical context that links related information leads to greater recall and clustering by reward category. The combined effect of learning rate and source features suggests that by strengthening the binding of items to their temporal features in addition to constructing de novo categories around value supports adaptive forms of organization.

Talk 12: Prediction under uncertainty: Dissociating sensory from cognitive expectations in highly uncertain musical contexts

IrisMencke, Max-Planck Institute for Empirical Aesthetics

Predictive models in the brain rely on the continuous extraction of regularities from the environment. These models are thought to be updated by novel information, as reflected in prediction error responses such as the mismatch negativity (MMN). However, although in real life individuals often face situations in which uncertainty prevails, it remains unclear whether and how predictive models emerge in high-uncertainty contexts. Recent research suggests that uncertainty affects the magnitude of MMN responses in the context of music listening. However, musical predictions are typically studied with MMN stimulation paradigms based on Western tonal music, which are characterized by relatively high predictability. Hence, we developed an MMN paradigm to investigate how the high uncertainty of atonal music modulates predictive processes as indexed by the MMN and behavior. Using MEG in a group of 20 subjects without musical training, we demonstrate that the magnetic MMN in response to pitch, intensity, timbre, and location deviants is evoked in both tonal and atonal melodies, with no significant differences between conditions. In contrast, in a separate behavioral experiment involving 39 non-musicians, participants detected pitch deviants more accurately and rated confidence higher in the tonal than in the atonal musical context. These results indicate that contextual tonal uncertainty modulates processing stages in which conscious awareness is involved, although deviants robustly elicit low-level pre-attentive responses such as the MMN. The achievement of robust MMN responses, despite high tonal uncertainty, is relevant for future studies comparing groups of listeners' MMN responses to increasingly ecological music stimuli

Talk 13: Novelty and uncertainty differentially drive exploration across development

Kate Nussenbaum, New York University

Across the lifespan, individuals frequently choose between exploiting options with known rewards or exploring unknown alternatives. While a large body of work has suggested that children may be more exploratory than adults, it is unclear how novelty and reward uncertainty differentially influence decision-making across age, since these two choice features are often correlated. Here, we used a version of a recently developed value-guided decision-making task (Cockburn et al., 2021) in a large sample of children, adolescents, and adults (ages 8 - 27 years, N = 122) to examine the separable influences of novelty and uncertainty on exploration across development. In line with prior studies, we found that exploration decreased across age. Critically, however, participants of all ages demonstrated a similar bias to select choice options with greater novelty. Decreases in exploration with increasing age were driven by stronger aversion to reward uncertainty in older participants. Reinforcement learning modeling revealed that while children simply inflated the utility of more novel options, adolescents and adults used novelty to buffer the influence of reward uncertainty (Cockburn et al., 2021), such that the utility of novel options was less influenced by uncertainty aversion. These findings suggest that, though often correlated, distinct features of lesser known choice options - novelty and uncertainty - differentially influence exploratory decision-making across development.

Talk 14: Neural representation of latent cause in credit assignment

Yanchang Zhang, University of California Davis

Humans have a remarkable capacity to make inferences based on structural knowledge, which may depend on the ability to assign credit for both experienced and inferred relationships, a process remaining less known. We scanned hungry participants (N=28) while they tracked two stimulus-reward systems for desserts, with each system comprising two stimuli of different visual categories with the same reward probabilities. We hypothesize that at feedback, the choice leading to an outcome reactivates in different subregions of occipitotemporal cortex; the underlying cause is reinstated in the orbitofrontal cortex (OFC) as common patterns across stimuli. Behavioral results from logistic regressions and Bayesian models show that participants learned to track both probabilities. Univariate activity in the OFC, ventromedial prefrontal cortex, and the hippocampus show robust effects (t(27)>4.3, p<0.0001) of belief confirmation measured by Kullback-Leibler divergence. We used multivariate pattern analyses (MVPA) to test for a reinstatement of the choice identity at feedback (cross-validated across runs). Left OFC and the hippocampus show a significant decoding accuracy (t(27)>3.745, p<0.001). To test if this reinstatement constitutes a reactivation of the identity representation from stimulus presentation, we performed MVPA training on the stimuli in separate forced trials and decoding choice identity at feedback in free choice trials. Significant decoding accuracy was found in the bilateral OFC and the amygdala (t(27)>4.6, p<0.0001). These findings support a model whereby causal choices are reinstated at feedback, coincident with prediction errors, to drive plasticity between co-active neural ensembles for the outcome and cause for learning.

Talk 15: Mnemonic Discrimination Ability Predicts Optimal Training Condition for Memory-Guided Inference Decisions

Noh Sharon, University of California, Irvine

Previous work has shown that one's ability to infer latent structure can predict their ability to engage in goal-directed planning and decision-making (Rmus, Ritz, Hunter, Bornstein, Shenhav, 2021). We examined whether this ability can be improved via targeted training interventions. Participants (N=122, 78 female, ages 18-89 years, additional data collection in progress) studied a series of object pairs presented during the learning phase. Unbeknownst to participants, the objects are associated with one another based on an underlying graph structure. For some participants (N=54), object pairs were split into 4 separate learning blocks to minimize memory interference of overlapping object pairs (blocked training), whereas for others (N=68), object pairs were randomly presented across the 4 learning blocks (randomized training). On a final test, participants were asked to judge the relative distances between two object pairs as a measure of structural inference ability. We reasoned that performance on this task should depend on the type of memory representations participants form during the learning phase. Therefore, participants also completed a memory test (Mnemonic Similarity Task, MST) to get a measure of individual differences in memory ability, specifically mnemonic discrimination. Interestingly, we found that MST scores interacted with training condition to predict performance on the judgment task: those with lower MST scores benefited from blocked training, whereas those with higher MST scores benefited from the randomized training, (F(1, 118) = 7.17, p = .008). These findings provide insight on how memory-guided decisions may be improved with individualized training interventions.


APRIL 23–26 • 2022

Latest from Twitter