Thursday 31 July 2014
Signatures Of Human Cognition
Electrocorticography (ECoG) is a method for in-vivo sampling of electrical currents from precisely localizable neuronal populations with a high sampling rate and excellent signal to noise ratio. Simultaneous recordings from a large number of brain sites make it possible to study the local activity of neuronal populations along with their dynamic interactions with other brain regions in real time. The speakers of the symposium will highlight some of the latest breakthroughs in ECoG research that have given rise to new waves of discoveries in the field of cognitive neuroscience. They will argue that the anatomical precision and temporal resolution of ECoG, and the simultaneous access it provides to distributed brain networks, make it a suitable method for decoding the electrophysiological signature of human cognition in experimental as well as natural conditions.
Josef Parvizi, Stanford University, USA
Jean-Philippe Lachaux, INSERM Lyon, France
Robert T. Knight, University of California Berkeley, USA
SYMPOSIUM 20: Multi-Frequency The Interplay Of Attention And Prediction In The Human Brain
Theoretical models and recent data suggest that the human brain is best viewed as a predictive machine. According to this view, brains learn by minimising the amount of prediction error, or surprise, caused by unexpected events. Neuroimaging and computational work have implicated several brain areas in feedforward propagation of predictions from higher- to lower-order regions and subsequent feedback of prediction errors from lower- to higher-order brain areas. Predictive coding models hypothesise that prediction errors are weighted by their precision, which ties into the notion of attention tuning. In this symposium, we will present a series of studies that support these ideas. Kok will show that attention can reverse the reduction of prediction errors that is observed in fMRI data when sensory inputs are predicted. Chennu will present results from intracranial and ERP recordings that demonstrate how attention can differentially modulate responses typically associated with probabilistic inference such as mismatch negativity (MMN), P300, and contingent negative variation (CNV). Garrido will show model-based connectivity evidence for a rightward attentional bias to unexpected events. Finally, Langdon will discuss the mechanisms by which attention modulates reward-based learning in a neural circuit model.
Peter Kok, Donders Institute for Brain, Cognition and Behaviour, Netherlands
Srivas Chennu, University of Cambridge, UK
Marta I. Garrido, University of Queensland, Australia
Angela J. Langdon, Princeton University, USA
Cognitive Modeling And Cognitive
Neuroscience: A Symbiotic Relationship
Cognitive modeling and cognitive neuroscience have traditionally been regarded as separate fields of study. Cognitive modelers infer underlying cognitive processes based on observable behavioral outcomes, while cognitive neuroscientists examine cognitive processes using neuroimaging measures. The emerging field of model-based neuroscience uses formal cognitive modeling to isolate specific cognitive processes and relate these to brain measurements to develop more fine-grained models of cognition. This approach allows for a reciprocal relationship between the fields of cognitive modeling and cognitive neuroscience that can both enhance our ability to make precise interpretations of patterns of brain activity, and also inform and constrain formal cognitive models based on brain measurements. This symposium highlights examples of the ways in which cognitive modeling and cognitive neuroscience can interact to address a broad range of questions. Birte Forstmann uses cognitive modeling to determine the necessity of basal ganglia structures in regulation of the speed-accuracy tradeoff. Alexander Provost describes a study in which neuroimaging and behavioral data are simultaneously modeled to examine spatial skill acquisition. Matthias Mittner presents a novel way of determining the specific cognitive processes that differentiate task-related from task-unrelated thoughts. Elise Mansfield explores the association between networks supporting adjustments in response caution and adaptive functioning outcomes.
Professor Birte Forstmann, University of Amsterdam, The Netherlands
Alexander Provost, University of Newcastle, Australia
Matthias Mittner, University of Amsterdam, The Netherlands
Renate Thienel, University of Newcastle, Australia
Cognition and Connectomics
Attempts to comprehensively map the constituent neural elements and interconnections of the brain—the so-called connectome—have spurred rapid advances in neuroimaging, with a plethora of methods now available for characterizing the macro-scale connectivity architecture of the entire cerebrum in unprecedented detail. These developments have caused a paradigm shift in cognitive neuroscience, with a major emphasis now being placed on understanding how cognition emerges from the functional integration of spatially distributed, functionally specialized neural systems. Traditionally, the bulk of imaging research into brain connectivity has focused on measurement of structural connectivity or functional interactions during task-free, so-called “resting-states”, although recent studies have begun to apply the tools of network science to map stimulus-evoked changes in brain functional network organization in order to understand the network determinants of cognitive processes. This symposium will provide an up-to-date introduction and overview of this field by examining basic concepts and techniques, their application to cognitive neuroscience experiments, and fundamental questions such as how brain function is constrained by network structure.
Luca Cocchi, Queensland Brain Institute, Australia
Andrew Zalesky, University of Melbourne, Melbourne, Australia.
Professor Olaf Sporns, Indiana University, USA
Michael W Cole, Rutgers University, USA
Formal Theories Of Dorsal Anterior Cingulate
Dorsal anterior cingulate cortex (dACC) is one of the most studied neural systems in cognitive neuroscience yet an understanding of its specific function remains elusive. Evidence from multiple experimental methodologies militates against the development of a unifying theory, implicating dACC in roles as various as conflict monitoring, motivation of effortful behaviors, task maintenance, error prediction, and more. In this symposium we will discuss recent computational modeling efforts to elucidate dACC function. Presentations will reflect a range of approaches including neurobiologically-inspired models that account for detailed neurophysiological data, and more abstract or normative models that focus on explaining functional neuroimaging data and on behavioral impairments following brain damage. The theories propose a variety of different functions for dACC – meta-learning (Khamassi), predicting the outcomes of actions and signaling discrepancies between observed and predicted events (Alexander), evaluation and specification of appropriate control (Shenhav), and hierarchical control over effortful behavior (Holroyd) – but share in common a central role for dACC in the adaptive regulation of behavior. Together these efforts hold out the promise of reconciling the divergent views of dACC function within a unifying theoretical framework.
William H. Alexander, Gent University, Belgium
Dr Mehdi Khamassi, CNRS Paris, France
Amitai Shenhav, Princeton University, USA
Clay Holroyd, University of Victoria, Canada
Do we make predictions from current sensory information about future sensory input? If so, what form to these predictions take? What are the brain mechanisms involved? Evidence that we do make predictions comes from wrong predictions: performance is reduced. There is a pattern of brain activity that is automatically recruited by prediction errors. What is not agreed is the form the predictions take—whether they simply involve down-regulation of the mechanisms, including memory mechanisms, encoding the current sensory input or whether there is an active process of constructing predicted inputs. The speakers will address these issues in audition and in vision. In both modalities regularities embedded in stimulus sequences are exploited to set up predictions about the what and the when of the forthcoming stimulus.
Robert P. O'Shea, Southern Cross University, Australia
Bradley N. Jack, Southern Cross University, Australia
Juanita Todd, University of Newcastle, Australia
Peter Keller, University of Western Sydney, Australia