Tuesday, 21 April 2015

Postcognitive Topics final post

During our post-cognitive topics module we have covered many different approaches to cognitive science from the Extended Mind to Enaction. Other approaches include Cognitive Neuroscience, Cognitive Neuropsychology and Computational Cognitive Science. One thing all the approaches have in common is a tendency to produce long papers - like this post.

Elman et al., in Rethinking Innateness, describe development as an interactive process where ‘emergent form is the rule rather than the exception.’ They go on to describe development as taking place at multiple levels and in discussing innateness say ‘development is constrained at one or more of these levels. Interactions may occur within and also across levels. And outcomes which are observed at one level may be produced by constraints which occur at another level.’ They are describing development in emergent terms.

Information Technology has ended up defining systems that in the end look like emergent systems. One system  in particular that demonstrates this is the seven layer Open Systems Interconnection (OSI) model of communications that starts at the electric impulses being sensed and ends up with full communication protocol for computers.    

The person working at the application layer uses completely different tools and processes than the person working at the physical layer. At the application layer you are dealing with users and other software applications. At the physical layer you are dealing with pins, voltages, cables and other hardware. It would never be expected to use the same approach or tools at different layers. A chip designer and an android developer could not swap jobs without retraining.

As Hutchins says in Enaction, Imagination and Insight, ‘the proper unit of analysis for cognition should not be set a priori, but should be responsive to the phenomena under study’. Similarly the proper approach to the analysis of cognition should not be set a priori, but should be responsive to the phenomena under study.  A pluralist framework is required to support this. Some way of evaluating outputs from the different approaches is required to justify the choice of approach. 

The Enactive approach, Froese and Di Paulo, outlines the emergence of new domains from different layers.  They outline a system which goes from cell to society. They recognise the different layers along the way. Each layer has its own constraints and processes and can support non-reducible domains of activity. New processes and domains emerge at each layer. Each layer uses affordances from its supporting layer. Each layer is dependent on the layers from which it emerges.

The point that Froese and Di Paulo do not cover is that emergence is agnostic and can support many different approaches. Research needs to be clear about the unit of analysis it is studying and the approaches being used. The unit of analysis of neuroscience is clearly measuring activity such as blood oxygen levels close to the neuron level. However it is working at a much different unit of analysis when it relates findings at this level to language or emotions. The mix of very different units of analyses with the same approach needs to be acknowledged and justified.

The advantage of emergence is that the different domains don’t need to be bound into the same processes and descriptions of supporting layers. For instance in Participatory Sense-Making An Enactive Approach to Social Cognition, Di Paolo poses the question as to how autonomy, agency, interaction and inter-subjectivity apply to the Enactive theory of social cognition. As social cognition is a different domain that emerges from individual cognition, it is not necessary to assume the same features and processes. Describing the interactions between social cognition and the individual cognition is essential. It is of interest to investigate how the constraints in supporting layers influence social cognition. Do inter-subjectivity or agency need to be stretched to apply to Social Cognition? Does Social Cognition need to be embodied? Do the processes and features at the cell level need to apply to all other levels? Pick an approach suitable to the unit of analysis on the basis that it adds value rather than on the strict adherence to one approach.

Postcognitive Topics postscript

This module exceeded my expectations. The readings and discussions were provocative and stimulating.  Highlights for me included how the passive dynamic walker can walk down a slope without any motor or control system in a most natural way; Runeson’s analysis of the planimeter allowing the appreciation that there is a logic in the bio'logical' that does not have to be the same as the logic with which we are familiar; von Uexküll and Umwelt Theory which reminded me of David Attenborough's work; and dynamic embodied situated cognition which I am starting to understand. Thanks Fred.

1 comment: