Cognitive Science and its critics

1
Cognitive Science and its critics

• Consciousness and Cognition
• 11. SECOND ORDER CYBERNETICS AND ENACTIVE
  PERCEPTION
• (Dr) Mark Bishop (Dr) Slawek Nasuto

2
Design for a brain

• It should be noted that from now on the system
  means not the nervous system but the whole
  complex of the organism and the environment.
  Thus, if it should be shown that the system has
  some property, it must not be assumed that this
  property is attributed to the nervous system it
  belongs to the whole and detailed examination
  may be necessary to ascertain the contributions
  of the separate parts.
• Design for a brain, (312 pp.41), W. Ross Ashby,
  (1952).

3
Introduction

• This seminar will highlight a problem - The
  External Observer Fallacy - inherent in any
  attempt to instantiate genuine cognitive systems
  within the framework of First Order Cybernetics
  and/or any flavour of Cognitivism/Computationalism
  .
• Computationalism cognition is instantiated by
  the execution of appropriate computations.
• In 2005 Bishop Nasuto argued that Cognition
  and Phenomenology (i.e. the subjective feel of
  sensation) in man, machine and animal are best
  understood within a new unified framework
  integrating Dynamic Systems Theory, Second Order
  Cybernetics and Enactive Perception.

4
Background

• Humans life exist within an ever-changing world.
• Classifying invariants, (cf. David Chalmers
  easy problem of consciousness)
• Bottom up theories of perception
• Top down theories of perception.
• Phenomenal perception of objects, (cf. David
  Chalmers hard problem of consciousness)
• The ineffable red of a rose.
• But how can we cross this seemingly an
  unbridgeable explanatory gap between Chalmers
  easy and hard problems?
• Perhaps one clue is that cognitive systems,
  (contra algorithms), are dynamic systems they
  are fundamentally embodied in time.

5
Dynamical systems theory

• Recall that a dynamic system is any system for
  which we can provide
• A finite number of state variables which capture
  the state of the system at a given time
• A set of state space evolution equations showing
  how the values of those variable change over
  time
• Parameters of the dynamic system affect the
  behaviour of the system without being modified
  themselves.
• In the phase space graph of the Lorenz attractor
  opposite (famous for demonstrating the butterfly
  effect) the state variables represent
  meteorological data.

Trajectory
6
Radical dynamic systems theory

• Cognitive systems are dynamic systems and are
  best understood from the perspective of dynamics.
• In viewing cognition simply as a continuous
  dynamic process, dynamicists explicitly reject
  the notion of cognition as the computational
  manipulation of internal representations.
• Thus the dynamic systems theory of cognition
  outlines how to intelligently interact with the
  world, without necessarily representing it.
• Further, the radical dynamic systems theory of
  cognition, (from Wheeler), simply conceptualises
  mental phenomena as state space evolution in
  appropriate forms of dynamical system.
• Instantiating appropriate system dynamics
  instantiates appropriate cognitive state, with
  concomitant mental phenomena located solely
  within the dynamic agent.

7
Post Cartesian ecologiescoupled dynamical
systems

• Coupling occurs when two separable dynamical
  systems are bound together and can affect each
  others parameter values.
• Changes in the first systems state values
  produce changes in the second systems parameter
  values to produce changes in the phase portrait
  of the second system
• and vice versa.
• As coupled systems can be described with one set
  of equations it is up to the observer as to where
  to draw the boundaries of the internal states of
  the system and the external states of the
  environment.

8
The entre-deux

• Hence in an ecological system the agent and
  environment can be seen, not as two distinct
  entities but one coupled dynamical system.
• The Cartesian approach does not account for the
  ongoing agent-environment interactions in this
  way.
• Merleau Pontys Entre-Deux, (Phenomenology of
  Perception Routledge pp. 430)
• The world is inseparable from the subject, but
  from a subject which is nothing but a project of
  the world
• and the subject is inseparable from the world,
  but from a world which the subject itself
  projects.

9
Cybernetics

• The Greek root of the word cybernetics -
  kybernetes can mean governor or steersman.
• Simple reflections about steering/control of a
  boat, yields many of the sub-disciplines that
  together define cybernetics.
• Cybernetic control fundamentally involves a
  feedback loop
• sensing the current course
• Study of sensors materials electronics etc.
• working out the course error
• Study of controllers (analogue and digital)
  comparing current course with desired course
  and establishing any necessary course changes.
• correcting acting upon the rudder to alter
  direction.
• Study of actuators.

10
First order cybernetics

• First order cybernetics is concerned with
  structures that are fundamentally de-coupled from
  their own construction by an engineer, (system
  designer / external observer).
• Hence there is a fundamental asymmetry between
  the engineer and any first order cybernetic
  system, (structure).
• The Engineer will design a structure to achieve
  behaviour consistent with proscribed purpose,
  (i.e. Its ascribed teleology).
• The Engineer can interpret the operation of
  such structures, (which may contain multiple
  complex feedback paths), in terms of
  teleological, (i.e. goal directed), behaviour.
• However any goal directed behaviour observed in
  the structure is - at some level - in reality
  ascribed, defined by the engineer.

11
The external observer fallacy

• The cognitive ability, (and/or teleological
  behaviour), ascribed to a first order cybernetic
  system is in reality a reflection of the
  cognitive abilities of the external observer.
• The external observer imposes a meaning on the
  observed behaviour of the (cognitive) system.
• This meaning that may not be unique
• as different observers may attribute different
  meanings to the system's behaviour.
• Consider the infamous Guardian advert where the
  same event was viewed by different people.
• The imposed meaning is not necessarily a true and
  intrinsic property/description of the systems
  operation.
• The explanation of the systems operation,
  ultimately, remains formal..

12
Second Order Cybernetics, (SOC)

• Although cybernetics means governor or steersman
• who or what steers the steersman?
• by what processes is the steersman steered?
• how does the steersman steer himself?
• Heinz von Foerster attributes the origin of
  second-order cybernetics to the attempts of
  classical first order cyberneticians to
  construct a model of the mind.
• von Foerster realized that
• a brain is required to write a theory of a
  brain. From this follows that a theory of the
  brain, that has any aspirations for completeness,
  has to account for the writing of this theory.
  And even more fascinating, the writer of this
  theory has to account for her or himself.
  Translated into the domain of cybernetics the
  cybernetician, by entering his own domain, has to
  account for his or her own activity. Cybernetics
  then becomes cybernetics of cybernetics, or
  second-order cybernetics.

13
Circular organisation

• Second Order Cybernetics is the cybernetics of
  systems involving their observers as opposed to
  the cybernetics of systems that are externally
  observed.
• Hence second order cybernetics fundamentally
  involves the observer as a constitutive part of a
  circular organisation.
• A circular organisation is best exemplified via
  the classic control problem of regulating room
  temperature using a heater and a thermostat.
• i.e. Does the thermostat control the temperature
  in the room or does the temperature in the room
  control the thermostat?

14
Brian Arthur and Complexity Theory

• It is ironic that one of the most elegant
  computer simulations of the core second-order
  cybernetic concept
• that models affect the very system they are
  supposed to model
• was not created by a Cyberneticist, but by the
  Economist/Complexity-Theorist Brian Arthur.
• Arthur simulated the seemingly chaotic behavior
  of stock exchange-like systems by programming
  agents that are continuously trying to model the
  future behavior of the system to which they
  belong, and use these predictions as the basis
  for their own actions.

15
Second Order Cybernetics and teleology

• Living systems are complex adaptive control
  systems engaged in circular relations with their
  environment
• Second order cybernetics recognises the absolute
  inseparability of the cognitive agent and the
  environment
• Teleological properties being emergent, not
  a-priori, properties of the combined
  agent-environment meta-system.
• Second order cybernetics encapsulates a
  constructivist epistemology
• Knowledge constructed only from experience
• i.e. agent / environment interactions.

16
Radical Constructivism

• Constructivism is the idea that knowledge cannot
  be passively absorbed from the environment, it
  must be actively constructed by the system
  itself
• It cannot be the result of a passive receiving
  but originates as the product of an active
  subjects activity.
• Radical constructivism, thus, is radical
  because it breaks with convention and develops a
  theory of knowledge in which knowledge does not
  reflect an objective ontological reality, but
  exclusively an ordering and organization of a
  world constituted by our experience. The radical
  constructivist has relinquished metaphysical
  realism once and for all
• (Ernst von Glasersfeld, 1981)

17
Second order cybernetics and radical
constructivism

• Second order cybernetic systems have no internal
  access as to how the world really is
• i.e. There is no homunculus observing an internal
  model of an external world
• Instead perceiving error-signals constantly
  indicate disturbance(s) from the system goal.
• In such SOC systems the only influence the world
  has on the system is to indicate when its model
  makes inaccurate predictions
• that means that the real world manifests
  itself exclusively there where our constructions
  break down, (von Glasersfeld).
• c.f. Heideggers, being-in-the-world and
  present-at-hand.

18
Inner models

• In a Second Order Cybernetic system inner
  models are not high fidelity representations of
  a pre-given outer-reality but subjective
  constructions that, by complex feedback loops,
  drive the system towards its goals.
• The danger of complete relativism, where any
  model is as good as another, is avoided by
  coherence and invariance.
• Coherence a social process whereby phenomena
  become real by consensus, (cf. the coherence
  theory of truth).
• Invariance the external world consists of
  entities which maintain their noumenonal form
  over time..

19
Second Order Cybernetics and Enactivism

• The principle of undifferentiated encoding from
  von Foerster (1980), tersely states, The
  response of a nerve cell encodes only the
  magnitude of its perturbation and not the
  physical nature of the perturbing agent.
• Ie. There is no difference between the type of
  signal transmitted from eye to brain or from ear
  to brain.
• This raises the question of how it is we come to
  experience a world that is differentiated, that
  has "qualia", sights, sounds, smells?
• The answer from SOC is that our experience is
  enacted it is the product of process
• Encodings or "representations" are interpreted as
  being meaningful (or conveying information)
  purely within in the context of the actions that
  give rise to them.
• What differentiates sight from hearing is the
  sensorimotor information that locates the source
  of the signal and places it in a particular
  action context, (cf. ORegan Noe).

20
ORegan and NoeA sensorimotor account of
vision

• Experience is not something we passively feel but
  something we actively do.
• Analogy is of driving a car.
• This automatically accounts for the difference
  between the different sensory modalities as each
  employ different sensorimotor skills.
• Cf. Muellers specific nerve energies - sensory
  pathways - and von Foesters principle of
  undifferentiated encoding
• Or specific cortical areas of activation why
  should sense modalities be different in different
  areas?

21
In vision

• No representations created, knowledge about the
  world is accessible via sensorimotor
  co-ordinated interactions with it.
• Use of the outside world as its own external
  memory.
• Consider the harmonica in the bag experiment.

22
Enactive perception

• Views the world as its own representation and
  vision as an embodied exploratory, (enactive),
  process of the world mediated by appropriate
  sensorimotor contingencies.
• Instead of regularly exploring the image opposite
  to build up a representation, we simply check to
  see it confirms our belief that it depicts a man
  and woman eating.
• We believe we maintain a rich visual
  representation of the world, because each time we
  enquire of some visual feature, our eye
  immediately saccades to attend to it.
• Hence enactive perception is also constructivist,
  as knowledge of the world is actively constructed
  by the perceiving agent through its interaction
  with the environment.
• i.e. Mental phenomena generated by richly coupled
  processes located within the meta-system of agent
  and environment interactions.

23
Some evidence for enactive perception

• In classical theories of vision, large scale
  differences in the two alternating images would
  cause very different patterns of activation in
  the brain.

24
Change blindness (1)

• The sensorimotor theory results in testable
  hypothesis relating to the phenomena of change
  blindness.
• Human change blindness suggests that our
  internal representations of the world are very
  sparse (no detail).

25
Change blindness (2)

• Consider the retinal blindspot where there are no
  photoreceptors yet we perceive a continuous - and
  rich - visual field.
• Similarly colour vision is weak outside the
  fovea centralis.

26
Change blindness (3)

• Even if we stare at the area of change in an
  image, the probability of detecting it is less
  than 60.

27
Change blindness (4)

• The further away a change takes place from the
  area that we are attending, the less likely we
  are to notice it.

28
Change blindness (5)
29
Change blindness (6)
30
Exploring the world

• No need for a rich internal representation of the
  world, we just need to learn algorithms for
  sensorimotor coordination to explore it.
• For example, in vision these algorithms would
  control the eye saccades as we attend visual
  features of the world.
• This process results in the grand illusion that
  we see the whole visual field albeit we actually
  only attend to a small part of it at any one time.

31
Seeing with our eye muscles

• Seeing only occurs when exorcising our
  sensorimotor mastery over a scene
• Image manipulation by moving our eyes.
• Only the aspects attended to by our eyes are
  actually perceived.
• In a famous flight simulator experiment two
  airline pilots out of eight continued to attempt
  to land their aircraft even though the runway had
  a small light aircraft parked across it.

32
The hard problem of qualia

• What is it about a red bit of paper that makes us
  experience red?
• Redness is defined by the processes that are
  carried out by the sensorimotor system when the
  changes of incoming reflectance spectra are
  typical of a red thing.
• Hence redness is not just the spectra of a point
  on a red surface but the relative changes we
  observe when looking across the scene.

33
Two views of (visual) perception

• Representation Vs Action Closing the gap

34
A unified view of cognition(1) Dynamic systems

• Cognitive systems are dynamic systems
  fundamentally enacted in time and best
  characterised and understood in the context of -
  and using the tools of - dynamic systems theory.
• Cognitive systems do not exist in isolation but
  are spatially embodied in an environment to form
  a coupled agent-environment system.

35
A unified view of cognition(2) Dynamic systems
SOC

• The coupled agent-environment system builds-up
  knowledge in a fundamentally second order
  cybernetic-constructivist manner.
• i.e. However the agent does not build-up a high
  fidelity representational model of an out-there
  world
• but instead the agent-environment interactions
  either confirm or disabuse the agent of veracity
  of its current model/knowledge and so the
  current model dynamically evolves over time.

36
A unified view of cognition(3) Dynamics SOC
Enactivism

• The enactive theory of perception perfectly
  extends the general framework of Dynamic Systems
  Theory and Second Order Cybernetics as it
  explicitly demonstrates cognitive processes as
  arising from the coupling of the observer and the
  environment.
• By explaining the inter-modal gap (e.g. why
  sight is like seeing and not like touch) and the
  intra-modal gap (e.g. why red looks red and not
  blue) the enactive theory of perception
  emphasises the role of sensorimotor coupling as
  the specific neuro-psychological mechanisms
  leading to the specific phenomenology of
  perception.

37
Conclusion

• In their 2005 paper in Kybernetes, Second order
  cybernetics and enactive perception, Bishop
  Nasuto argue that all first order cybernetic (
  cognitivist) approaches to cognition ultimately
  share the same fate the external observer
  fallacy
• and suggest that the best conceptual framework
  to avoid this fallacy is to integrate Dynamic
  Systems, Second Order Cybernetics and the
  Enactive Theory of Perception.