Centre for Cognitive Science (COGS)

Summer term 2010

  • Week 1 (Apr 20th) Dr Mike Beaton (COGS, Sussex): What is Conceptualism?
  • Week 2 (Apr 27th) Prof Steve Torrance (COGS, Sussex): Ethics and Artifice
  • Week 3 (May 4th) Dr Robert Clowes (COGS, Sussex; and University of Lisbon): Mind against the background of Social and Cognitive Technology
  • Week 4 (May 11th) Prof Axel Cleeremans (Université Libre de Bruxelles): The Reach of the Unconscious
  • Week 5 (May 18th) Dr Pasha Parpia (COGS, Sussex): Reappraisal of the Somatosensory Homunculus and its Discontinuities
  • Week 6 (May 25th) Dr Tony Morse (University of Plymouth): Thinking with your body & developmental cognitive robotics
  • Week 7 (Jun 1st) No Meeting
  • Week 8 (Jun 8th) Dr Tom Froese (Sackler Centre/COGS/Informatics, Sussex): Validating and calibrating first-person methods
  • Week 9 (Jun 15th) Dr Thomas Nowotny (RCUK Academic Fellow, Informatics, University of Sussex): A Parallel Implementation of a Biologically Realistic Spiking Neuronal Network Model of Unsupervised Olfactory Learning on Graphical Processing Units.
  • Week 10 (Jun 22nd) No meeting

Week 1

Date: Tuesday, 20th April 2010
 Dr Mike Beaton (COGS, University of Sussex)

What is Conceptualism

Conceptualism is the thesis that a subject can only perceive the world to be a certain way, if the subject has those concepts required to understand what it is for the world to be that way. I will argue in favour of this thesis, and will also try to argue that it is entirely compatible with the enactive, embodied and anti-representationalist turns in cognitive science. Conceptualism is not a thesis about language, nor is it a thesis about internal symbol processing. Rather, it is a thesis about the kinds of things an agent must be able to do, in order to perceive. I this talk, I will review a way of thinking about concepts in terms of whole-agent level rational action. I will suggest that John McDowell has had many important insights about the nature of rationality. However, McDowell thinks that conceptualism is incompatible with animals minds. I will argue that he is wrong on this, and that we can (indeed, must) acknowledge a kind of non-reflective, entry-level rationality, which is in operation in us and in animals. Finally, I will present a way of understanding why concept exercise, on this view, fundamentally involves the external world. I will suggest that this means that at the most fundamental level, perceiving and acting do not involve internal representations of what is perceived and acted upon.

Week 2

Date: Tuesday, 27th April 2010
 Prof Steve Torrance (COGS, University of Sussex)

Ethics and Artifice

As with AI proper, the field of Artificial Ethical Intelligence (AEI) presents a number of challenges. Artificial agents are getting more and more functionally autonomous, so it might look like a good idea to provide them with abilities to make 'correct' moral judgments - but is it? But could an AEI research programme provide us with anything like a full-blooded artificial ethical agent? If only rather impoverished approximations are possible then might artificial moral agents be worse than useless in real live situations? Also, is ethical intelligence all there is to ethical agency, or are there affective, phenomenological, social or other factors which are perhaps less amenable to AI methods? And: under what conditions could there be artificial ethical 'patients' (or recipients) to which we owe some kind of moral consideration, as opposed to ethical 'agents' from which we expect morally responsible choices?

Many of these questions have been implicit in AI from the beginning, but now they have to be addressed with more focus and clarity as AI enters into the age of Artificial Ethics. This talk is based on a plenary presentation at the AISB-10 conference, April 2010.

Week 3

Date: Tuesday, 4th May 2010
 Dr Robert Clowes (COGS, University of Sussex; and University of Lisbon)

Mind against the background of Social and Cognitive Technology

Human minds are exquisitely porous and open to the world and have a particularly interesting relationship to technology. Social and cognitive technologies such a Wikipedia, Google Search, iPhones, Facebook have caused euphoria and consternation amongst commentators and some dark reflections on what it is to be human. But can we view these technologies against a wider background of humanity's incorporation of technology into our life-worlds and into ourselves?

Should these new technologies be regarded as challenges to our humanity, or new ways of actualizing it?

This talk is an expanded version of one Dr Clowes gave to the RSA (Royal Society for the Arts and Commerce).

Rob Clowes is Tutorial Fellow at COGS: the Centre for Research in Cognitive Science at the University of Sussex and is a Research Fellow at the New University of Lisbon. He is currently completing his book Being Human After Facebook.

Week 4

Date: Tuesday, 11th May 2010
 Prof Axel Cleeremans (Université Libre de Bruxelles)

The Reach of the Unconscious

A great conceptual pendulum oscillates, with a period of about 30 or 40 years, over our understanding of the relationships between conscious and unconscious information processing. Its path delineates the contours of the unconscious mind as well as its contents: Sometimes smart and defining the very fabric of the mind, the unconscious is at other times relegated to taking care of little more than our bodily functions. At this point in time, the pendulum finds itself hovering rather steadily on the side of those who think so many functions are served by the unconscious that they even question the necessity of consciousness.

Here I will suggest that the pendulum has swung a little too far, and illustrate the argument with recent experimental findings that document how challenging it may be to arrive at a satisfactory conception of the relationships between conscious and unconscious information processing. I will focus on two recent studies, one concerning visual awareness, and the other on decision-making. Both are suggestive that the specific methods we use, as well as the manner in which we interpret the data, are of profound importance with respect to the conclusions we draw about the power of the unconscious.

Week 5

Date: Tuesday, 18th May 2010
 Dr Pasha Parpia (COGS, University of Sussex)

Reappraisal of the Somatosensory Homunculus and its Discontinuities

Neuroscience folklore has it that there are two discontinuities in the spatial-adjacency mapping between the touch receptors in the skin and neurons in human primary somatosensory cortex (two discontinuities in somatotopy). In Penfield's homuncular map (above), the hands and face, although not spatially adjacent in the body, map onto adjacent regions in cortex. The same applies to feet and genitalia. It has been proposed that these conjunctions in cortex result from coincident sources of stimulation in the fetal position, where the hands frequently touch the face, and the feet the genitalia (Farah, 1998). Computer modelling using a Hebbian variant of the self-organizing Kohonen net is consistent with this proposal.

I will present recent literature that reveals that the genital representation of touch (as opposed to tumescence) in primary cortex is continuous with that of the lower trunk and thigh. In this respect Penfield and coworkers were in error. Furthermore, studies of somatosensory maps in Thalidomide patients and others born with missing or severely foreshortened arms do indeed exhibit normal hand-face adjacency in cortex.

These results, in conjunction anatomical findings of separate face innervation and its earlier onset of sensory function, compared to that of the rest of the body, allow a reappraisal of homuncular organization. I propose that the somatosensory homunculus comprises two distinct somatotopic regions: the face representation and that of the rest of the body. Principles of self-organization, sometimes glibly applied by computational modellers of brain function, do not account satisfactorily for the overall homuncular map.

Week 6

Date: Tuesday, 25th May 2010
 Dr Tony Morse (University of Plymouth)

Thinking with your body & developmental cognitive robotics

In this talk I will present an overview of the iCub robot, of what developmental cognitive robotics is and why it is important, and then in more detail discuss a series of developmental cognitive robotics experiments in which the iCub robot is used to replicate psychology experiments on spatial biases in categorisation. These experiments highlight a much broader impact of embodiment than merely offloading computation into the body and environment or using the body to constrain problem space. While the importance of embodiment in cognitive theories is gaining ground, we have only just begun to understand the implications of embodiment on our conception of cognition.

Week 7

Date: Tuesday, 1st June 2010
 No Meeting

Week 8

Date: Tuesday, 8th June 2010
 Dr Tom Froese (Sackler Centre/COGS/Informatics, Sussex)

Validating and calibrating first-person methods

After over a century of neglect the last two decades have seen a significant amount of progress in the science of consciousness. This resurgence of interest has been largely driven by the availability of increasingly sophisticated neuroscientific methods. However, as the field is maturing it is becoming progressively more evident that further scientific progress will not only depend on improvements in measurement technology. Additionally, there are two major outstanding challenges that need to be addressed. We still need a theory of consciousness that could inform the design and interpretation of experimental studies. And we also need a more systematic way of accessing and measuring the phenomenology of consciousness, i.e. our lived experience. The latter challenge takes a special place because a rigorous method of obtaining phenomenological data may turn out to be a powerful catalyst for the field as a whole. Only with increasingly refined verbal reports about what it is like to be conscious can we hope to better understand the detailed data that neuroscience is providing, and to delimit the phenomenological facts that a theory of consciousness must take into account.

Accordingly, it is not surprising that there have been a growing number of attempts to go beyond employing standard questionnaires and spontaneous verbal reports of untrained subjects in the science of consciousness. I will describe some of the outstanding examples of how more rigorous first- and second-person methods can inform empirical research. However, there are conflicting claims between the two most prominent second-person interview methods regarding the existence and accessibility of previously unattended experience. I discuss the possibility of a methodological resolution of this conflict, which is based on ideas produced by a recent working group within the Sackler for Consciousness Science focused on this topic. It may be possible to objectively validate and calibrate first- and second-person methods of consciousness.

Week 9

Date: Tuesday, 15th June 2010
 Dr Thomas Nowotny (RCUK Academic Fellow, Informatics, University of Sussex)

A Parallel Implementation of a Biologically Realistic Spiking Neuronal Network Model of Unsupervised Olfactory Learning on Graphical Processing Units.

Parallel computing has been used for a long time, at least since the late 50s and early 60s. However, only recently has parallel computing grown beyond the domain of expensive super-computers and entered the broader market in the form of multi-core processors. Somewhat independently, and less noticed by the general scientific community, graphic processing units (GPUs) also have become powerful, highly parallel computing devices which can now challenge the de-facto monopoly of the few big CPU manufacturers.

In this talk I will present the parallel implementation of a spiking neuronal network model with biologically realistic morphology, elements, and function on a GPU using the NVidia CUDA framework. The model describes a prototypical olfactory system of converging and diverging pathways that performs unsupervised odor recognition (clustering) using a spike timing dependent plasticity (STDP) learning rule.

When comparing the parallel implementation of the model to a well-designed standard C/C++ implementation I observed a 24x speedup when using an NVidia Tesla C870 device for the CUDA implementation and a 3 GHz AMD Phenom II X4 940 processor for the classical implementation. With this speedup, the CUDA program can run the model comprising 2670 neurons and on the order of 200,000 synapses in faster than real time.

The work I present is one of a handful of published proof of concept studies of GPU computing in computational neuroscience but I expect the field to expand exponentially in the near future. The new technology holds the promise of overcoming some of the limitations large scale modeling has faced in the past.

Week 10

Date: Tuesday, 22nd June 2010
No Meeting

Series organized by Prof. Steve Torrance (stevet@sussex.ac.uk)