Roundtable talk recordings

Abstract: The overarching goal of my research is to understand how we recognise different visual objects in the environment, with a specific focus on the recognition of socially relevant signals. Our remarkable ability to “read the room” is a form of social intelligence that emerges during infancy and contributes to our social wellbeing, yet its neural basis is only partially understood. How do know when a stranger is looking directly at us from across the street? How do we track changes in someone’s mood during a conversation (and can we do this efficiently via zoom)? How do we (can we) ignore social cues during criminal proceedings? To address these questions and others, I combine psychophysics with state-of-the-art neuroscientific methods (including whole brain functional imaging, single-cell recordings and inactivation techniques) and I test multiple species, including humans and rhesus macaques. In this RoundTable discussion, I will describe recent empirical work investigating (1) the neural correlates of face pareidolia and other unusual faces, (2) the link between perception of naturalistic, spontaneous facial expressions and brain activity and (3) the difference between interaction and observation modes in the primate brain. This work sets the stage for future experiments that will help us understand how, as social primates, we are able to communicate and coordinate with other social agents.

Bio: Jess Taubert was awarded her PhD in Psychology from the University of Sydney in 2009. Her first postdoctoral position was in the laboratory of Lisa Parr (Emory University, USA) where she spent her days comparing the behavioral responses of chimpanzees and rhesus macaques and trying to understand how our closest living primate relatives recognize faces. In 2011, she accepted a fellowship to move to Belgium (KU Leuven) and record single unit responses in fMRI-defined regions of the macaque brain (supervised by Rufin Vogels, Wim Vanduffel, and Bruno Rossion). Later, Jess found her way back to the University of Sydney to work with David Alais and Frans Verstraten. During this time she drove investigations about temporal effects in perceptual judgements about faces (and wrote papers about the dangers of online dating platforms like Tinder). Then in 2016, Jess took a position as a senior research fellow in the laboratory of Brain and Cognition, headed by Leslie Ungerleider (NIMH, USA). At NIMH Jess studied the brain mechanisms underlying the detection of illusory faces in primates. In 2020 she was awarded the ARC Future Fellowship and moved to the University of Queensland in August 2021 where her lab is busy trying to understand the processes underlying human social intelligence and their dysfunction. Fun fact: Jess’ first experiments investigating face perception during her honours year, compared human behavior with the behavior of domestic chickens.

Email Website

Abstract: Accurate decisions tend to be both confident and fast. Nonetheless, there are relatively few models that can simultaneously address this three-way relationship, especially for single stage decisions where participants indicate both their choice and their confidence. Extending on a common decision architecture of the linear ballistic accumulator framework, two models have been proposed – 1) a Multiple Threshold Race model which instantiates the Balance-of-Evidence hypothesis where confidence is determined through the difference between accumulated evidence for competing options (e.g., Reynolds, Osth, Kvam, & Heathcote, in revision), and 2) a newly developed Confidence Accumulator model which assumes that confidence itself is accumulated independently for each confidence option. To test these two confidence architectures, we ran two experiments manipulating the length of the confidence rating scale across 2-, 4-, or 6-options in a recognition memory task along with a perceptual task. Different models were compared that made different allowance for how the length of the confidence scale affected model parameters. While both model classes found that thresholds were affected by the length of the scale, drift rates were only minimally affected. Implications for models of confidence and response time will be discussed.

Bio: Haomin Chen graduated from the University of Melbourne with a B.A. majoring in Psychology in 2019, and obtained a 1st Class Honours in Psychology in 2020. Haomin is currently a 3rd year PhD student working under the supervision of Dr. Adam Osth in the Melbourne School of Psychological Sciences at the University of Melbourne. Her research is focused on investigating the three-way relationship between confidence, response latency and accuracy using a decision model.

Email

Abstract: Numerous studies have found that the Bayesian framework, which formulates the optimal integration of the knowledge of the world (i.e. prior) and current sensory evidence (i.e. likelihood), captures human behaviours sufficiently well. However, there are debates regarding whether humans use precise but cognitively demanding Bayesian computations for behaviours. Across two studies, we trained participants to estimate hidden locations of a target drawn from priors with different levels of uncertainty. In each trial, scattered dots provided noisy likelihood information about the target location. Participants showed that they learned the priors and combined prior and likelihood information to infer target locations in a Bayes-fashion. We then introduced a transfer condition presenting a trained prior and a likelihood that have never been put together during training. How well participants integrate this novel likelihood with their learned prior is an indicator of whether participants perform Bayesian computations. In one study, participants experienced the newly introduced likelihood, which was paired with a different prior, during training. Participants changed likelihood weighting following expected directions although the degrees of change were significantly lower than Bayes-optimal predictions. In another group, the novel likelihoods were never used during training. We found people integrated a new likelihood within (interpolation) better than the one outside (extrapolation) the range of their previous learning experience and were quantitatively Bayes-suboptimal. We replicated the findings of both studies in a validation dataset. Our results showed that Bayesian behaviours may not always be achieved by a full Bayesian computation. Future studies can apply our approach in different tasks to enhance the understanding of decision-making mechanisms.

Bio: Sophie is a postdoctoral researcher in Marta Garrido's Cognitive Neuroscience and Computational Neuropsychiatry Lab at the University of Melbourne, developing the first wearable magnetoencephalography (MEG) in Australia and investigating decision-making using the Bayesian framework. Previously, she worked as a research fellow in 2017-2019, under the supervision of Chris Miall (University of Birmingham) and Gareth Barnes (University College London), applying wearable MEG to investigate learning signals in the human cerebellum. She did her PhD in computational neuroscience at Imperial College London. Supervised by Aldo Faisal, she used Bayesian decision theory and psychophysics to relate increased sensorimotor uncertainty to a higher risk of fall incidents in older people. Not by coincidence, she is also a neurologist trained in Taiwan.

Email Website

Abstract: Background: Blindsight refers to the uncanny ability of cortically blind patients to respond correctly, or above chance level, to visual stimuli, despite having lesions in the primary visual cortex. Affective blindsight more specifically refers to the ability of correctly identifying emotional expressions presented to the patients' blind fields. This suggests that there may be alternate pathways for visual information processing in the brain, which can bypass the primary visual cortex and reach other areas that are involved in visual perception. One theory of affective blindsight posits that the amygdala, plays a key role in this phenomenon while another theory proposes that the phenomenon of affective blindsight can be explained by functional integrity of the extra-striate visual areas, via the projections from the superior colliculus and pulvinar. The posterior superior temporal sulcus (pSTS) is an extrastiate brain region that is involved in the processing of social and emotional information, such as facial expressions, gestures, and body language. Current research suggests that the pSTS may play a role in affective blindsight. However, there is limited literate on this. We report a rare case of affective blindsight for in whom we demonstrate a preserved pSTS with evidence for connectivity to pulvinar that bypass the primary visual cortex. Methods: A 58 year old man developed sudden onset of seeing flashes of light across his entire visual filed that lasted for a few seconds. This was immediately followed by complete blackness and loss of vision. He presented to us 6months later with persistent visual loss. Based clinical evaluation and structural MRI he was found to have bilateral occipital lobe infarcts involving the primary visual cortices. On testing vision, he had no phenomenological vision in the central field but had a peripherally located island of vision in the right inferior visual field. Psychophysical testing, lesion mapping, lesion network mapping and resting state fMRI, were performed. The patient's structural MRI was spatially normalized to the MNI space, and the lesion location was manually mapped. We then used resting-state functional connectivity from a healthy cohort to identify the network of regions functionally connected to the lesion site. For this the lesion map was used as seed in a resting state functional connectivity analysis and a statistically thresholded map was defined as the lesion network. Results: Psychophysical testing revealed objective evidence of blindsight for color and facial expression recognition based on the 2-alternate forced choice paradigm. Lesion mapping showed that the lesion involved the region of the occipital face area(OFA). Lesion network mapping shows that the lesion network extended additionally to involve the fusiform face area(FFA) but spared the pSTS . Functional conncectivity analysis showed intact direct connectivity of the pulvinar nuclei to the pSTS without involving the lesioned lateral geniculate body to visual cortex pathway. There was no significant functional connectivity between the amygdala and pSTS. Conclusions: In conclusion, our study adds to the literature of affective blindsight by demonstrating a role for the pSTS and its connections to the subcortical pathways via the pulvinar. As there is strong evidence for the pSTS in processing facial expressions, its preservation and its intact connections to the pulvinar are likely to be critical in the development of affective blindsight. The preserved island of vision were peripherally located in the visual field and are therefore unlikely to contribute to affective blindsight in our patient as the stimuli was presented in the central and blind fields. We did not find any significant connections between the amygdala and the pSTS in our study to support the amygdala theory of affective blindsight. We also show that connections from primary visual cortex to OFA and FFA are not necessary for affective blind sight.

Bio: Dr A.T. Prabhakar is a neurologist who is currently a professor of neurology in the department of neurological sciences, Christian medical college (CMC) Vellore.  He obtained his undergraduate and post graduate medical training from CMC Vellore. He joined as faculty in 2013 after completing his D.M in clinical neurology at CMC, Vellore. He was a visiting professor at the Montreal neurological institute Canada in 2017 and trained in lesion-symptom mapping. His areas of interest include clinical phenomenology, cognitive Neuroscience, Lesion based studies in cognitive  neuroscience,  CNS inflammatory disorders,  visual neuroscience, Consciousness studies, neuro-philosophy and Critical Care.

Email Website

Abstract: The use of technologies such as social media and video games has reshaped human perceptions, actions, and environments and has been associated with a myriad of benefits for individuals. Nevertheless, it started revealing several “dark sides” with adverse consequences. In this talk, Prof Ofir Turel will present findings from multiple studies conducted over the last 8 years, attempting to uncover the neural and psychological-behavioral underpinnings of problematic social media behaviors. Such problematic behaviors include excessive use (sometimes called “addiction”), and other unsafe and risky online behaviors (e.g., sexting, revealing private information online, taking selfies in dangerous places, and using social media while driving). He also discusses the emergent picture, as well as several open research questions.

Bio: Ofir Turel is Professor of Information Systems Management within the School of Computing and Information Systems at The University of Melbourne. He is also a Scholar in Residence at the Decision Neuroscience Program, University of Southern California. His research interests include a broad range of behavioral-psychological, managerial, and bio-physiological issues related to the use and management of computer information systems and technologies. He has published papers in top business outlets, as well as in leading psychology, psychiatry and neuroscience journals. Example psychiatry and neuroscience outlets include Journal of Psychiatric Research, Cognitive Affective & Behavioral Neuroscience, Social Neuroscience, Behavioural Brain Research, Brain Imaging and Behavior, Neuroscience Letters, Progress in Neuropsychopharmacology & Biological Psychiatry, Psychiatry Research, Biological Psychiatry, and Addiction Biology.

Email Website

Abstract: We use quantities in everyday life to guide our decisions: we judge the size and number of vegetables at the market; we dance following the tempo of music; we compare how many steps we need to reach our daily goal. How are we able to flexibly interact with such different quantities in our environment? Ultra-high resolution (7 Tesla) fMRI combined with population receptive field modelling has revealed a widespread network of brain areas that selectively process quantities, such as numerosity (object number) and object size, which are organised as topographic maps. These maps are distinct but overlap with visual field maps representing orientation and eccentricity. In this talk, I will discuss four of my recent projects that extend this line of work and illustrate (1) a network of topographic maps in association cortex that hierarchically transforms visual timing-selective responses, (2) local image contrast representations in early visual cortex are transformed into location-independent numerosity tuning in extrastriate cortex, (3) a similar derivation follows for visual timing-tuned neural responses from early visual stimulus representations, and (4) the extent of overlap and topographic direction between maps for numerosity and visual event timing supports the view of a supramodal functional network for cognitive quantities that facilitates cross-modal interactions. I will conclude by considering the implications of a supramodal functional network for learning to map quantities to symbols.

Bio: Dr Jacob Paul is a cognitive neuroscientist at the Melbourne School of Psychological Sciences. His interests include visual perception, numerical cognition, and the neurocognitive development of maths reasoning. His research spans multiple levels of explanation from modelling longitudinal patterns of maths learning, to characterising individual differences in numerical decision making, and mapping the organisation of neural circuits underlying transformations of visual quantities. He has recently returned to Melbourne after postdoctoral work in the Utrecht (Netherlands) combining ultra-high resolution 7T fMRI and computational modelling to characterise how the brain encodes basic quantities like number and time. He is passionate about translating his research findings to help enhance functional numeracy levels in the population, improve learning and instruction of mathematics, as well as constructing brain-based interventions and remediation programs for individuals with maths learning difficulties.

Email Website

Abstract: We present a descriptive model of choice with normative foundations based on how the brain is thought to represent value. An individual’s behavior is fully described by two primitives: an individual’s expectation and one free parameter we call “predisposition’. The model captures the same apparent preference phenomena illustrated by Prospect Theory but unlike Prospect Theory accounts for individual heterogeneity in parameters, employs far fewer parameters than full prospect theory, and retains neurobiological plausibility as a causal model of the choice process. Additionally, our theory makes a series of novel predictions amenable to future testing and includes an alternative explanation for endowment effect.

Bio: Prof Tymula (University of Sydney) combines theory and methodology from economics, psychology, and neuroscience to understand how people decide, why they make wrong decisions, and how to make them better choosers. Her research focused on how economic preferences change over the lifespan and contexts, including how thirst, being observed, outdoor luminance, as well as the structure of current and past choice sets affect behaviour. Prof. Tymula has been awarded over $33 million in grants as a Chief Investigator, including ARC Centre of Excellence, DECRA, Discovery, and Linkage grants. In 2017 she received the Award from the Society for Neuroeconomics for her contributions to our understanding of decision-making. She is a board member of the Society for Neuroeconomics and Executive Committee Member of the Economic Science Association.

Email Website

Abstract: The human brain must integrate information across the two hemispheres of the brain to construct a coherent representation of the perceptual world. Characterising how visual information from the left and right visual fields is represented in the two hemispheres (ipsilateral/ contralateral) over time can be the missing piece in understanding how information is transferred between hemispheres. In this talk, I will discuss how we can use electroencephalography and computational methods to investigate information processing within and across the hemispheres. I will present results that provide new insights into the dynamics of object perception and the competitive versus cooperative nature of hemispheric processing.

Bio: Dr Amanda Robinson is an ARC DECRA Fellow at the University of Sydney with expertise in perception and attention, specifically relating to object perception. She received her PhD from the Queensland Brain Institute at The University of Queensland for research on olfactory-visual integration. After a postdoc with Prof Marlene Behrmann at Carnegie Mellon University, Dr Robinson returned to Australia to join the Sydney Computational Cognitive Neuroscience Lab. Dr Robinson uses EEG, MEG and fMRI with computational methods to uncover the spatiotemporal dynamics of perceptual processing.

Email Website

Abstract: It is widely assumed that we must use predictions to determine the nature of our perceptual experiences, but theories within the domain of action are at odds with those from other perceptual disciplines. Specifically, action theories propose that we cancel predicted events from perception to render our experiences informative - telling us what we did not already know. In contrast, theories outside of action - typically couched within Bayesian frameworks - demonstrate that we combine our predictions (prior) with the evidence (likelihood) to determine the experience (posterior). Such a process would render our experiences more veridical in the face of sensory noise. I will present several psychophysical and neuroimaging findings from our lab that ask whether action predictions really influence perceptual processing so differently. I will conclude by asking how we may, in principle, use predictions to generate experiences that are both veridical and informative.

Bio: Clare undertook her PhD at the UCL Psychology department with Cecilia Heyes and Patrick Haggard, before undertaking three fellowships. The first was again with Celia at UCL, the second with Martin Eimer at Birkbeck and the third at the Wellcome Centre for Human Neuroimaging, UCL, with James Kilner and Karl Friston. Her first faculty position was at the University of Reading, before moving to her current position at Birkbeck in 2012 where she runs the Action Lab. All questions relate to action, perception, and learning, and most commonly the links between those domains.

Email Website

Abstract

Individual differences in almost all complex traits arise from processes involving both genes (G) and the environment (E). To analyze and deepen our understanding of such GE interplay is one of the major challenges at the research frontier today and essential if we wish to identify environmental factors that have true causal effects on complex traits. In this talk, I will provide some examples from my work on expertise and skill acquisition using music as a model behavior, highlighting how we can apply methods using large scale twin and genetically informative data to disentangle GE interplay and strengthen causal inferences.

Speaker Bio

Miriam A. Mosing is a Senior Researcher and DRM Fellow at the Melbourne School of Psychological Science, and an Associate Professor at the Department of Medical Epidemiology and Biostatistics at the Karolinska Institute in Stockholm, Sweden. Her lab investigates (1) expertise development and (2) quality of life throughout lifetime and in the aged, using interdisciplinary approaches to quantify the interplay between genes and the environment. She is involved in a range of international consortia exploring genetic factors underlying complex traits, including the Interplay of Genes and Environment across Multiple Studies (IGEMS) consortium, The Loneliness Consortium (TLC), GENEtic research into Quality Of Life (GENEQOL), and the Musicality Genomics Consortium (MusicGenes) among others. Her research is currently funded by a NIH RO1 and a Wallenberg grant to explore gene-environment interplay underlying SES health gradients in late life and health effects of active engagement in culture, respectively.

Email Website

Abstract

To navigate the world safely, it is critical that we are able to rapidly evaluate and change our mind about our perceptual decisions. For example, imagine being unable to overrule a decision to cross the street when you realise a speeding car is approaching. In situations such as this, even small delays in the time it takes for you to change your mind can have serious consequences. In this talk, PhD Student, William Turner, presents findings from three studies which examined the cognitive processes underlying changes of mind (Studies 1-2) and retrospective evaluations of decision confidence (Study 3). Study 1 found that the likelihood and speed of decision reversals depends on both relative and absolute sources of sensory information, while Study 2 found that even the very earliest sensory information one receives influences later change-of-mind decisions. These results challenge existing computational models of the processes underlying changes of mind. Will presents new variants of these models which better account for these findings, and outline ways in which their assumptions can be tested further. Finally, Study 3 shows that information which is seemingly extraneous to a perceptual decision—specifically the amount of effort invested into reporting a decision outcome—nevertheless influences retrospective judgements of decision confidence. This suggests that humans are sensitive to a ‘motoric sunk cost effect’ whereby forgone physical effort expenditure can inflate decision confidence.

Speaker Bio

William Turner is a PhD candidate in the Melbourne School of Psychological Sciences. He is interested in the cognitive and neural processes underlying perceptual decisions and decision reversals.

Email Website

Abstract

Humans routinely learn the value of actions by assessing their outcomes. Critically, actions also require effort. While effort is known to affect the valuation of rewards, how effort modulates reward-based learning remains unclear. Here, we applied a reinforcement learning paradigm in which individuals (N = 213) had to exert predefined levels of force to register their responses. Our key finding was that greater exertion of effort boosts teaching signals that are positive, but blunts those that are negative. Moreover, the extent to which effort reinforced learning for a given individual was proportional to their aversion to effort. This suggests that the same computation that discounts value before choice serves to reinforce learning after that choice is made. Overall, our findings are consistent with growing evidence that motivation and learning operate within a common computational framework, and refine current reinforcement learning models.

Speaker Bio

Huw Jarvis is a PhD student in the Cognitive Neurology Lab at the Turner Institute for Brain and Mental Health at Monash University. Prior to his PhD, he completed studies in medical science (University of Tasmania) and public health (University of Melbourne), and worked in research translation at the National Health and Medical Research Council. Huw is undertaking a PhD to investigate the computational underpinnings of motivation and reward-based learning, with a view to better understanding related impairments in neurological and psychiatric disease. He was recently awarded a Fulbright Future Scholarship to pursue related work at Yale University.

Email Website