Workshop overview

This is a small exploratory workshop on the topic of predictive coding models of mental and neural function. Numbers are limited so please email the organizer first if you are not yet included and wish to attend. The workshop brings together perspectives from Neuroscience, Philosophy, Informatics, Music, and Anthropology, and opens with a tutorial session by Professor Karl Friston, Scientific Director of The Wellcome Trust Centre for Neuroimaging, UCL.

The workshop is funded by the ESF/AHRC CONTACT project, with the support of the Department of Philosophy, the School of Philosophy, Psychology and Language Sciences, and the School of Informatics.

Organizer - Andy Clark: andy.clark@ed.ac.uk

Programme

All sessions to be held in the Informatics Forum, room G.07A

19th January 2010

9.30	Coffee
9.45-11.00	Tutorial Session: The Bayesian brain, surprise and free-energy Professor Karl Friston (Scientific Director of The Wellcome Trust Centre for Neuroimaging, UCL) Chair: Andy Clark (Edinburgh)
11.00-11.15	Break
11.15-12.15	Prediction error, rubber hands, and the skin Dr Jakob Hohwy (School of Philosophy & Bioethics, Monash University, Australia) Chair: Liz Irvine (Edinburgh)
12.15-13.30	Lunch
13.30-14.30	Shared Affective Motion Experience (SAME) and Minimised Prediction Error (MPE) Dr Katie Overy (Music, Institute for Music and Human Development, University of Edinburgh) Chair: Julian Kiverstein (Edinburgh)
14.30-14.45	Break
14.45-15.45	Unsupervised probabilistic models as normative models of the visual system Professor Chris Williams (Director of the Institute for Adaptive and Neural Computation and Professor of Machine Learning, Informatics, University of Edinburgh) Chair: Matteo Colombo (Edinburgh)
15.45-16.45	Bayesian machines in interaction: surprising, adapting and coupling Professor Andreas Roepstorff (Center of Functionally Integrative Neuroscience, Aarhus University, Denmark) Chair: Tillman Vierkant (Edinburgh)
16.45-17.00	Break
17.00-18.00	Discussion Session: Free Energy and Prediction - A View from Cognitive Science Dr Chris Thornton (Department of Informatics, University of Sussex) Chair: Mog Stapleton (Edinburgh)

 

Note: Professor Karl Friston will also be giving a Cognitive Neuroimaging Seminar on "Attractors in Song" on Monday 18th at 5pm in the Gaddum Lecture Theatre, 1 George Square.

Abstracts

Professor Karl Friston

The Bayesian brain, surprise and free-energy

Value-learning and perceptual learning have been an important focus over the past decade, attracting the concerted attention of experimental psychologists, neurobiologists and the machine learning community. Despite some formal connections; e.g., the role of prediction error in optimizing some function of sensory states, both fields have developed their own rhetoric and postulates. In work, we show that perception is, literally, an integral part of value learning; in the sense that it is necessary to integrate out dependencies on the inferred causes of sensory information. This enables the value of sensory trajectories to be optimized through action. Furthermore, we show that acting to optimize value and perception are two aspects of exactly the same principle; namely the minimization of a quantity (free energy) that bounds the probability of sensations, given a particular agent or phenotype. This principle can be derived, in a straightforward way, from the very existence of biological agents, by considering the probabilistic behaviour of an ensemble of agents belonging to the same class. Put simply, we sample the world to maximise the evidence for our existence.

Dr Jakob Hohwy

Prediction error, rubber hands, and the skin

The idea that the brain represents by minimising prediction error is very powerful. It can impact on philosophy by casting a new light on philosophically relevant areas of psychology. One such area is the study of the rubber hand illusions and here I bring some core properties of the prediction error minimisation notion to bear on this illusion. I present data from work done with Bryan Paton and discuss how prediction error, and the Bayesian approach it comes with, challenges some common ideas about the body and agency. I also present data that seems to suggest that the body itself, rather than just the brain, is an instrument at the service of prediction error minimisation.

Dr Katie Overy

Shared Affective Motion Experience (SAME) and Minimised Prediction Error (MPE)

Abstract to come

Professor Chris Williams

Unsupervised probabilistic models as normative models of the visual system

In this talk I will discuss the work by a number of authors (e.g. Olshausen and Field, Bell and Sejnowski, Karklin and Lewicki, Rao and Ballard, Hyvarinen and co-authors) on unsupervised probabilistic models as normative models of V1 and higher levels, and explain why I believe such methods will play a crucial part in understanding sensory processing.

Dr Andreas Roepstorff

Bayesian machines in interaction: surprising, adapting and coupling

We learn from the Bayesianists that the brain is all about prediction, e.g. the causes of events, the consequences of actions or, anticipated perceptions. But what happens when the percept is the actions of another person? We examined that in experiments on the perception and production of tapping done with Peter Vuust and Ivana Konvalinka. In single-person passive listening paradigmed, we observed, not surprisingly, MEG responses to aberant events that could be interpreted as error signals and competence dependent neural resolutions. In joint tapping experiments, we identified behavioural signatures of mutual adaption that suggested a coupling of the two individuals. I will present these findings and discuss whether some of the same mechanisms can be generalised also to exchange of higher order tokens, such as symbols and objects.

Dr Chris Thornton

Free Energy and Prediction: A View from Cognitive Science

The free-energy framework has interesting implications for issues in cognitive science relating to knowledge and representation. But there are also some puzzles about how those implications should be interpreted. The talk will present a selection of these for discussion.

Useful links

• Travelling to Edinburgh
• University of Edinburgh Campus Maps
• Accommodation in Edinburgh (Tourist Board)
• Accommodation in Edinburgh (Casamundo)

 

 

---

Last updated: January 11th 2010 by Mog Stapleton.