Spring term 2011 : Previous COGS seminars : ... : Centre for Cognitive Science (COGS) : AV��Ƶ

Week 1 (Mon. Jan 10th) - Antonio Chella (University of Palermo): Conceptual Spaces for Robot Perception and Introspection
Week 2 (Tues. Jan 18th) - Special Session: Steve Torrance, Robert Clowes, Blay Whitby, Mike Beaton: Social implications of a future intelligence explosion
Week 3 (Jan 25th) - David Leavens (AV��Ƶ): Idols of the Theatre: False Demonstrations of Human Uniqueness in Gestural Signalling
Week 4 (Feb 1st) - Julie Coultas (AV��Ƶ): Experimental methods and cultural evolution research: The case of the cultural transmission of disgust
Week 5 (Feb 8th) - Chris Thornton (AV��Ƶ): Poached Eggs, Black Ravens and Grue: Is there any need to mention Induction?
Week 6 (Feb 15th) - Ricardo Sanz (Universidad Politécnica de Madrid): Ten Design Rules for a Conscious System
Week 7 (Feb 22nd) - Hugo Critchley (BSMS, Sackler Centre): How bodily arousal impacts on the content of consciousness: Insights from neuroimaging
Week 8 (Mar 1st) - David Gamez (Imperial College): The Scientific Study of Human and Machine Consciousness
Week 9 (Mar 8th) - Kim Bard (University of Portsmouth): A Developmentally-Based Comparative Perspective on the Role of Emotional Engagements in Social Cognition
Week 10 (Mar 15th) - No Talk
Week 11 (Mar 23nd) - Prof Olga Pollatos (University of Potsdam, Germany): The interaction of interoceptive awareness on emotion regulation and empathy for pain

Week 1

Date: Monday 10 January 2011
Speaker: Prof. Antonio Chella

Abstract: Conceptual Spaces for Robot Perception and Introspection

We refer to a cognitive architecture for robot vision that has been developed during more than 10 years at the RoboticsLab of the University of Palermo. The architecture is organized in three computational areas: the subconceptual, the conceptual and the linguistic area. The subconceptual area is concerned with the low level processing of perceptual data coming from the sensors. In the linguistic area, representation and processing are based on a logic formalism. Representations in the conceptual area are couched in terms of a conceptual space that provides a principled way for relating high level, linguistic formalisms on the one hand, with low level, unstructured representation of data on the other.

We propose that robot perceptions are the perceptions of the outer world, described by means of suitable points in first-order conceptual spaces. To model higher order perceptions in introspective agents, we introduce the notion of higher-order conceptual spaces. Each point in these spaces corresponds to a self-reflective agent, i.e., the robot itself, people, and other robots with introspective capabilities.

Short Bio

Antonio Chella was born in Florence, Italy, on March 4, 1961. He received the Laurea degree in electronic engineering and the Ph.D. degree in computer engineering from the University of Palermo, Italy, in 1988 and 1993, respectively. Currently, he is a Professor of Robotics at the University of Palermo and the Head of the Robotics Lab. In 2008 and 2009 he was a member of the Core Program Committee of the AAAI Fall Symposium on BICA (Biologically Inspired Cognitive Architectures). In 2007 organized and co-chaired the AAAI Fall Symposium on AI and Consciousness. He served as a reviewer for EU Commission and for Italian and French Governments for projects related with robotics and cognitive systems. He has been the coordinator of the course of studies of "Computer Engineering" of the University of Palermo from 2001 to 2007. He has been the Director of the Department of Computer Engineering from 2007 to 2010. He is an Associate Editor of the Artificial Intelligence Journal. He is founder and Editor-in-Chief of the "International Journal of Machine Consciousness". He is a Founding member of the BICA Society. He authored or co-authored more than 200 papers concerning robotics, cognitive systems, machine consciousness, neural networks.

Week 2

Date: 18 January 2011
Speaker: Prof. Steve Torrance
Responses: Dr. Robert Clowes, Dr. Blay Whitby, Dr. Mike Beaton

Abstract: Social implications of a future intelligence explosion.

A growing group of informed opinion considers the hypothesis of a proliferation of super-smart AIs as worthy of serious scientific and philosophical attention. Ideas such as the Technological Singularity, Transhumanism (or Humanity+), the Superintelligence Explosion, Artificial General Intelligence (AGI), etc. - previously considered as on the academic fringe - are now being taken more seriously by researchers in AI, computing and philosophy.

What would be the social implications of any future explosion of super-AI agents? We will try to make a first pass on that question by looking at three particular areas: Would such super-AIs be conscious (in the phenomenal, 'what-it-is-like' sense)? Should they be considered as general moral agents (either as morally responsible beings or as beings worthy of moral concern)? Would they become genuine participants in our social and political frameworks (or us in theirs)?

Format of the session:
Presentation (20 mins): Steve Torrance;
Responses (10 mins each): Robert Clowes, Blay Whitby, Mike Beaton.
Followed by a floor discussion.

Week 3

Date: 25 January 2011
Speaker: Dr. David Leavens

Abstract: Idols of the Theatre: False Demonstrations of Human Uniqueness in Gestural Signalling

Numerous claims have been made for human uniqueness in signaling characteristics, even in non-verbal and pre-verbal contexts. Here I will outline some of the more common errors in reasoning that have led researchers to characterize human non-linguistic patterns of communication as biologically distinct from those of our nearest living relatives. In his seminal, The New Instrument, Sir Francis Bacon warned of false notions that "so beset men's minds that truth can hardly find entrance" (1). One class of false notions that he identified were those that peculiarly afflict scientists:

. . . there are Idols which have immigrated into men's minds from the various dogmas of philosophies, and also from wrong laws of demonstration. These I call Idols of the Theater, because in my judgment all the received systems are but so many stage plays, representing worlds of their own creation after an unreal and scenic fashion. Nor is it only of the systems now in vogue, or only of the ancient sects and philosophies, that I speak; for many more plays of the same kind may yet be composed and in like artificial manner set forth; seeing that errors the most widely different have nevertheless causes for the most part alike. Neither again do I mean this only of entire systems, but also of many principles and axioms in science, which by tradition, credulity, and negligence have come to be received.

The contemporary study of gestures across species exemplifies Bacon's prediction: It is a grand stage play founded upon obviously false axioms (philosophical dogma) and inept methodological approaches (wrong laws of demonstration). An example of the first type of error is the presumption that behavioral innovation is a consequence of evolutionary history: examples are legion in which human children display new competencies and these competencies are attributed to the children's evolutionary, rather than ontogenetic histories. The typical onset of pointing by children in Western cultures at around 12 months of age has been almost universally interpreted by scientists as a biological adaptation for referential communication, as though nothing that happens in a child's first 12 months of life is relevant for understanding how pointing is manifested in human development. An example of the second type of error is the widespread practice of comparing gesture use across species, when the individual organisms used were raised in completely different circumstances from birth and typically tested at incommensurate stages of life.

Week 4

Date: 1 Feb 2011
Speaker: Dr. Julie Coultas

Abstract: Experimental methods and cultural evolution research: The case of the cultural transmission of disgust

In this talk I will begin by briefly describing cultural evolution research. Theories about cultural evolution have been around since the 1980s (Cavalli-Sforza & Feldman, 1981, Lumsden and Wilson, 1981, Boyd and Richerson, 1985). However, it is only recently that we have seen a proliferation of experiments on cultural transmission (e.g Philosophical Transactions of the Royal Society B, 2008, 363). I will talk briefly about hypothesis testing and cultural transmission experiments using Boyd and Richerson's conformist transmission model as an example. I will then focus on recent research methodology in cultural transmission; illustrating it with some recent work that I have undertaken with Kimmo Eriksson at the Centre for Cultural Evolution, Stockholm. Much experimental work on cultural transmission, including the serial reproduction method, has taken the 'traditional' psychology route and used university undergraduates as participants. There has been criticism of this (Kashima and Yeung, 2010) for lacking external and ecological validity. I will talk about our current work on the cultural transmission of disgust in stories and our recent data collection methods: (1) Amazon Mechanical Turk and (2) cultural transmission 'in the wild'.

Week 5

Date: 8 Feb 2011
Speaker: Dr. Chris Thornton

Abstract: Poached Eggs, Black Ravens and Grue: Is there any need to mention Induction?

Induction is the process by which seen data becomes the basis for prediction of unseen data. Believed to be fundamental at various levels of cognition, it is also the process by which scientific knowledge is produced. There has long been a desire to understand its general principles. But fundamental difficulties stand in the way. Hume's circularity problem and the no-free-lunch theorems both seem to suggest the logical impossibility of a general mechanism. Faced with these difficulties, Machine Learning takes the position that there is no general mechanism. But an interesting alternative comes from Epistemology. Popper's falsificationist theory holds that there is a general mechanism, but that it is not *performing* induction. Inductive effects arise implicitly, through pursuit of a non-inductive goal. Less plausibly, the underlying mechanism is taken to be uninformed exploration of hypotheses. But as the present paper shows, Popper's solution can be reworked using information theory. Instead of taking inductive effects to arise from hypothesis search, they can be taken to arise from enhancement of representations. A mathematical argument can be used to show that increasing the informational efficiency (i.e., mean symbol content) of representations produces inductive effects. Representation enhancement can then take the place of hypothesis-search in the argument. The paper sets out this reformulation and considers the degree to which it prepares the ground for a general reduction of machine learning to information theory.

Brief Bio

Dr. Chris Thornton is a lecturer in the Department of Informatics at the AV��Ƶ. In the previous departmental incarnation at Sussex, he was a lecturer in the School of Cognitive and Computing Sciences (COGS). Before that he was a lecturer in the Department of Artificial Intelligence at the University of Edinburgh. Chris's research focusses on use of information theory to develop and extend machine learning techniques. Previous projects have targeted data compression, cognitive science, neural networks, computational creativity, the problem of induction, kolmogorov complexity, sequence prediction, algorithmic music, analogy formation, minimum description length encoding, information refinement and formal concept analysis.

Week 6

Date: 15 Feb 2011
Speaker: Prof. Ricardo Sanz

Abstract: Ten Design Rules for a Conscious System

The quest for the universal controller technology -a technology for building machine minds for any purpose- has advanced by small, both practical and theoretical steps but without a clear convergence into a unified view. However, recent developments in search of improvements in open-environment robustness have produced a reactivation of the quest for the very essence of the mental -from a systemic/cybernetic perspective. This talk will present an architecture-centric proposal for a fundamental control structure that fulfills a basic set of requirements for being an explanation of a functional mind (including associated concepts such as percepts, knowledge, thinking, action, etc.). This structure is grounded on systemic, embedded control systems concepts so as to be realizable in machines. This proposal goes from the elementary aspects of sensing and perception to the higher aspects of knowledge, meaning and consciousness. The proposed design rules will provide a basic stance for understanding access consciousness and self-consciousness and a catalog of design features needed for the engineering of a conscious system of technological and economical value.

Speaker Bio

Ricardo Sanz was born in Tomellosa de Tajuña, Spain, in 1963. He is an electrical engineer -with a control systems focus- and doctor in robotics and artificial intelligence with a "Premio Extraordinario de Tesis" for his PhD thesis on strategic control architectures for complex, large-scale, continuous processes. He is professor in Automatic Control and Systems Engineering at the Universidad Politécnica de Madrid in Spain and Visiting Research Fellow at the AV��Ƶ, Sackler Center for Consciousness Science. For many years he has been involved in research projects in the field of complex intelligent control systems; ranging from expert system application to continuous process industries or software platforms for distributed intelligent embedded systems, to reverse engineering rat-brains or adaptive cognitive architectures for human-like agents. He is the coordinator of the UPM Autonomous Systems Laboratory (www.aslab.upm.es), a small research group with a simple but ambitious research program on the development of universal technology for autonomy engineering. He usually acts as expert for the European Commission and the Spanish Government research agencies in the fields of real-time embedded systems, automatic control, dependability, software-intensive systems, complex systems and cognitive systems. He is associated editor of the recently created International Journal of Machine Consciousness. His current research topics are focused around universal cognitive architectures -general mind designs for animals and machines- capable of robust autonomous behaviour. His main interest, as an engineer, is the applicability of these to dependable technical systems (e.g. factories, cars, utilities, etc.) but he doesn't neglect the possibility of explaining biological minds up to the level of human consciousness. In particular he is currently interested in the way cognitive systems can implement self-awareness loops to augment their functionality and robustness.

Week 7

Date: 22 Feb 2011
Speaker: Prof. Hugo Critchley

Abstract: How bodily arousal impacts on the content of consciousness: Insights from neuroimaging

States of bodily arousal are coupled to states of attention and effort. Feedback about one's internal physiological state can influence thoughts and feelings. I will present a set of behavioural and neuroimaging studies examining cardiovascular arousal and its impact on early processing & subsequent appraisal of sensory stimuli. These suggest levels at which cardiovascular arousal might block or enhance the processing of sensory information with differential effects on stimuli evoking distinct emotions.

Brief Bio

Professor Critchley's long term interest is the control of emotional and motivational behaviour. He now focuses on examining psychophysiological mechanisms underlying symptom expression in physical and psychological disorders; i.e. how brain and body interact to influence subjective experience, behaviour and physical health. Techniques typically combine functional brain imaging with autonomic monitoring and clinical studies of relevant patients. This work extends into behavioural medicine and is supported by the Wellcome trust.

He is also a member of the Sackler Centre for Consciousness Science (SCCS) faculty. The SCCS is a joint venture between the Schools of Informatics and Psychology at the AV��Ƶ, and the Brighton and Sussex Medical School. Its remit is to unravel the complex neural networks underpinning conscious experience, in health and in disease, and its foundation has been made possible by a generous donation from the Dr. Mortimer and Theresa Sackler Foundation.

Week 8

Date: 1 Mar 2011
Speaker: Dr. David Gamez

Abstract: The Scientific Study of Human and Machine Consciousness

The talk will start by showing that the virtual reality model of perception is the only coherent way of explaining how phenomenal experiences can appear outside of the body. The virtual reality framework also accounts for hallucinations, dreams, and out of body experiences, and it leads to a clear distinction between the phenomenal world of our experiences and the physical world described by science. This distinction will be used to dissolve the hard problem of consciousness and to demonstrate how a science of consciousness is possible. Such a science will be based on mathematical and algorithmic theories of consciousness that can make precise predictions about human conscious states, and it will be validated by comparison with first person reports. The last part of the talk will explain how this approach to the science of consciousness relates to work on machine consciousness. Initially artificial systems will support work on human consciousness by operating as test beds for the development of algorithmic theories of consciousness. Once an empirically grounded theory of human consciousness has been established, we will be able to use it to make convincing predictions about the actual consciousness of artificial systems.

Brief Bio

David Gamez studied for a BA in natural sciences and philosophy at Trinity College, Cambridge, and completed a PhD in Continental philosophy at the University of Essex. After a couple of years working on agent-based artificial intelligence he completed a second PhD on machine consciousness as part of an EPSRC-funded project to build a conscious robot. He is currently at Imperial College, London, where he is working on new techniques for the simulation and analysis of spiking neural networks. Gamez is the author of What We Can Never Know - a book exploring the limits of philosophy and science through studies of perception, time, madness and knowledge - and the co-editor of What Philosophy Is - a collection of essays on the nature of philosophy. He is currently working on a book that will provide a framework for the scientific study of human and machine consciousness.

Week 9

Date: 8 Mar 2011
Speaker: Prof. Kim Bard

Abstract: A Developmentally-Based Comparative Perspective on the Role of Emotional Engagements in Social Cognition

Comparative studies of social cognition often lack appropriate consideration of development, in general, and of the interaction of early rearing environment with development. In this talk, I present evidence that emotional engagements with social partners, with objects, and with social partner's affect about objects are crucial in the development of joint attention (JA) in chimpanzees, and each are influenced by previous interactive history. Interestingly, joint attention commonly occurred, without specific training, in 8-, 9-, and 10-mo nursery-reared chimpanzees (n=35) during standardized cognitive testing. Using hierarchical multiple regression analysis to predict JA success, I found that emotion accounted for a significant, F(2,28) = 7.75, p<.002, and large (36%) percentage of variance in JA success. Sociability (to the examiner and to the favourite caregiver) contributed a unique 15% improvement in prediction of JA success, F(2,26) = 3.96, p<.03. When compared at the same age, a group of Chester Zoo chimpanzee infants and a group of human infants from Cameroon were found to initiate joint attention equally often. But positive emotion from social partners surrounding JA events occurred significantly more often in the human group than the chimpanzee group. I conclude that developmental histories of emotional engagements determine social cognition and emotional experiences must be considered in comparative studies of joint attention.

Speaker Bio

KIM A. BARD is Professor of Comparative Developmental Psychology, Director of the Centre for the Study of Emotion at the University of Portsmouth, UK, and President of the Primate Society of Great Britain (see www.psgb.org ). Prior to arriving at University of Portsmouth in 1999, she was Research Scientist at Yerkes National Primate Research Center of Emory University, where she investigated the roles of emotion and socialization in early development, and designed a Responsive Care Nursery for chimpanzees to enhance their species-typical development. She received her BA with Honors in Psychology from Wheaton College, and her PhD in Comparative /Developmental Psychology from Georgia State University, based on fieldwork with orangutans in Borneo, Indonesia.

Kim Bard has a distinctive comparative and developmental perspective, which concerns understanding the evolution of developmental processes. She conducts empirical studies with an eye to clarifying universal and species-specific characteristics of great apes and humans. She has documented the similarity between chimpanzees and humans in the development of primary intersubjectivity (e.g., neonatal imitation, mutual gaze, socio-emotional communicative expressiveness), of early social cognition (e.g., social referencing), and of self-recognition.

Prof. Bard has more than 60 peer-reviewed publications and 30 book chapters, and has held grants from NIH-NICHD, NIH-NCRR, FP6-ICT, British Council, and The Leverhulme Trust. She currently serves on the Science, Research, & Practice Advisory Board of the Down Syndrome Educational Trust, UK, on the Advisory Board of Primates and is Associate Editor of the British Journal of Psychology.

Week 10

No Talk

Week 11 (NB After Term)

Date: Weds 23 Mar 2011
Speaker: Prof Olga Pollatos
Location: BSMS Medical Research Building (MRB) - Ground floor meeting room
This talk is jointly organised with BSMS (the Brighton and Sussex Medical School)
Note non-standard day of week (WEDNESDAY), and non-standard venue for this talk: This talk will be held in BSMS Medical Research Building, Ground floor meeting room (room G19) - this is in building 45 on this map: http://www.sussex.ac.uk/aboutus/findus/documents/campusmap.pdf

Abstract: The interaction of interoceptive awareness on emotion regulation and empathy for pain

Mental processes related to visceral activity have regained growing interest during the last few years. Bodily changes and the ability to perceive these bodily changes ("interoception", "interoceptive awareness") have found their way to modern neuropsychological theories such as the somatic marker theory of Damasio. It incorporates feedback from the periphery to the cortex and postulates that many mental processes are influenced by "marker" signals arising through bioregulatory processes. Perceiving internal signals and their changes interacts with emotion processing and could facilitate decision making and performance in many everyday situations. A process closely related to the accessibility of internal signals is emotion regulation. The ability to cognitively regulate emotional responses to aversive events is important for our mental and physical health. We investigated reappraisal as one emotion regulation strategy in participants in relationship to interoceptive awareness.

Participants attentively watched neutral and negative pictures while suppressing negative affect by means of reappraisal. High density EEG and spatio-temporal current density reconstruction were used to characterize the time course of emotion regulation and to identify brain regions involved in suppressing negative affect. Reappraisal was accompanied by reduced arousal and significant modulation of electrocortical activity during suppressing negative affect. Current density reconstruction revealed that interoceptive awareness modulates activation in both first- and second-order representation areas like the insula, the somatosensory cortices and the anterior cingulate. Preliminary data will also be presented on the interaction between empathy and thereby an EEG study using an empathy for pain paradigm and interoceptive awareness.

Speaker Bio

Olga Pollatos is professor of psychology at the University of Potsdam, Germany. Her research focuses on affective neuroscience and the relationship between emotional processes and bodily signals. Personality traits like alexithymia, depression and emotional coping styles are of interest and are assessed in interaction with their underlying neural correlates examined by high density EEG or functional MRT. Further research deals with the interaction between cognitive processes and bodily arousal as measured by heart rate, electrodermal activity or breathing. She is also interested in the perception and processing of pain and empathy for pain. Other research interests refer to the interaction of time perception and internal signal processing. The interest on the interface to psychology and psychiatry and neurology extends the research field into eating disorders, somatoform and anxiety disorders and neurological movement disorders.

Olga Pollatos studied psychology and medicine at the technical university of Berlin and the Ludwig-Maximilians-University of Munich. Her PhD on cardioception and emotion processing was absolved in Munich in the Biological Psychology section. Her venia legendi in psychology was achieved at the Ludwig-Maximilians-University of Munich in 2008. Her medical dissertation was located in the neuroradiology on olfactory processing and emotions. Short term visiting positions included stays at the university of Nottingham and the university of San Diego. She is professor at the university of Potsdam since 2009.