UNIVERSITY OF HERTFORDSHIRE
             COMPUTER SCIENCE RESEARCH COLLOQUIUM

                          presents

"Do Enactive Cognitive Robots Need a Self? Lessons from Neuroscience"

                     Dr. Elena Antonova
(Department of Psychology, Institute of Psychiatry, King's College London)
                            and
                  Prof. Chrystopher L. Nehaniv
    (Adaptive Systems Research Group, University of Hertfordshire)

                      8 May 2012  (Tuesday)
                         11 am - 12 noon+


                Hatfield, College Lane Campus
              * * * Lecture Theatre E351 * * *

                Everyone is Welcome to Attend

                Refreshments will be available

Abstract.

Efforts in Artificial Intelligence, in particular and most promisingly,
AI Enactive Cognitive Robotics seek to produce life-like agents by
emulating the capacities seen in nature, particularly in humans.
Human experience appears to involve several distinct notions of 'self',
including somatic self, narrative self, social self, and experiential
self. These 'selves', arising in ontogeny, are best thought of as
processes that serve diverse and useful functions in humans.  Their
dynamic interplay and flexibility in shifting from one to another in a
context-dependent manner is essential for adaptive (healthy) behaviour
and cognition. The loss of such flexibility in humans can lead to various
psychopathologies, such as depression and schizophrenia. Common to most
human psychopathologies is the predominance of the narrative self-process,
which on the psychological level is experienced as a self-entity separate
from the other/environment. The narrative self has a distinct functional
neural correlate in the human brain, the Default Mode Network (DMN), which
runs in the midline of the brain, and minimally includes medial prefrontal
cortex and the posterior cingulate. The ability to modulate and attenuate
the DMN in humans is associated with greater behavioural flexibility,
cognitive efficiency, emotional regulation, and general well-being.
   Increasingly, AI robotics designers are finding they need to include
mechanisms for narrative and social intelligence, introducing self-models,
or targeting 'self-awareness'. In the light of this, it seems that
careful assessment is needed in determining what types of self-processes
an enactive cognitive robot should embody and whether some modes of
operating self-processes might reduce rather than increase functional
and 'metabolic' efficiency.
   In conclusion, in humans and hence, for similar reasons, in future
enactive cognitive robots, 'self' is best conceived of as an
interplay of transitory processes arising in a context-dependent
and a function-specific manner. 'Self' as a continuously running
process/module/network 'overseeing' the job of other processes in a
top-down manner could be functionally inefficient and even detrimental.


---------------------------------------------------
Hertfordshire Computer Science Research Colloquium
http://cs-colloq.stca.herts.ac.uk