Friday, 22 February 2008

Wynne, 2001

A summary of "Universal Plotkinism: A Review of Henry Plotkin's Darwin Machines and the Nature of Knowledge, by C. D. L. Wynne, 2001"


This is a summary of a review of a book. A researcher from the University of Western Australia wrote the review. It is encouraging to know that there are other researchers in Australia interested in this area.

Plotkin, 1993 is a book that contributes to the area of evolutionary epistemology and universal Darwinism. Evolutionary epistemology is "the idea that evolution is a knowledge gaining process." Universal Darwinism is similar to the general view of evolution with a larger scope that includes adaptation at the individual learning and cultural learning levels.

Plotkin, 1993 offers approaches for looking at questions such as following: "Why do some organisms learn?", "How did learning evolve?" and "How does behaviour relate to evolution?"

Plotkin's view that knowledge is a form of adaptation offers a useful perspective to analyse the relationship between evolution and knowledge. From an artificial intelligence perspective, it links research on animal behaviour with possible approaches to the development of an artificial intelligence. In particular, from an embodied artificial intelligence perspective Plotkin, 1993 offers intriguing avenues for consideration because the embodied approach, at a superficial level seems to bring concepts to the table whose value is not obvious. The embodied concept of the body being important for intelligence raises such questions as, "Why are some animals more intelligent when they all have bodies?"

To this question and others in the same vein, Plotkin, 1993 contributes by offering mentalist ideas to behaviourist models. These models have struggled with questions such as, "What is the origin of behaviour and what is its role in evolution?", "Why do some animals learn?", "What are the limits to learning?" and "What is the relation between evolution and culture?"

Wynne's review seeks to introduce Plotkin, 1993 to a behaviourist audience and to identify areas in Plotkin, 1993 that require further rigour, such as the sections on universal Darwinism.


Plotkin's thesis can be summarised in three statements. Firstly, that:

"...evolution can be characterised as a process, independent of the structures and mechanisms in which that process is instantiated."

Secondly, that:

"...this process is an epistemological or knowledge-gaining one."

Thirdly, that:

"...the oldest or most basic form of the evolutionary process is instantiated in genes and phenotypes. This "primary heuristic" has spawned other systems - in particular individual learning, immune responsiveness and cultural transmission of knowledge - that operate according to the same process and are constrained and informed by the primary heuristic to be both knowledge gaining and adaptive."

Plotkin's first statement that the mechanism in which evolution occurs being unimportant, supports the idea of universal Darwinism. Dawkins, 1983 coined the term "universal Darwinism", but the idea has been around much longer.

In regards to Plotkin's second statement, Wynne, 2001, presented the following examples to demonstrate that adaptations are knowledge. The first example is that a Martian can deduce much about earth by looking at an earthling's eyes. The second example is that plant life reflects the conditions surrounding them such as how the thin leaves of the Australian eucalyptus reflects the need to save water.

Plotkin holds that biological adaptation is knowledge for two reasons. Firstly, natural selection generates knowledge. Secondly, learning mechanisms use similar natural selection processes. To quote Plotkin, 1993, "When we come to know something we have performed an act that is as biological as when we digest something."

Plotkin's third statement is that there is a nested hierarchy of selectionist processes. When the primary heuristic is insufficient, the secondary heuristic of adapting by gaining knowledge occurs.

Plotkin holds that the primary heuristic both drives and constrains the secondary heuristic. This is important because it means that one does not begin life with a blank slate.

Plotkin holds that the secondary heuristic may lead to the third heuristic, which is knowledge exchange or in other words, culture.

[This post is a snippet from a longer summary.]


This section seeks place Plotkin, 1993, and Wynne, 2001, in the context of the research topic, natural language emergence in embodied artificial intelligence societies in virtual environments.

Plotkin, 1993 discusses ideas about evolutionary epistemology and universal Darwinism. Evolutionary epistemology relates to the research topic because it is about the origin and development of knowledge. Universal Darwinism relates to the research topic because of the idea of evolution being independent of the material that it occurs.

Plotkin's, 1993 can be distilled into three ideas, that evolution happens at the genetic level, the personal learning level and the cultural learning level.

The idea about how evolutionary techniques underlie individual learning reminds one of contrast between embodied and disembodied approaches. Embodied approaches seek to develop mechanisms to generate the act. Disembodied approaches seek to use descriptions of an act as a mechanism to generate the act. For example, some grammars are descriptions of natural language communication and in themselves do not necessarily generate natural language communication, yet disembodied approaches seek to use grammars to general natural language communications.

The ideas in Plotkin, 1993 are interesting and have equivalents in embodied artificial intelligence literature. It may be unwise to seek to borrow ideas from evolutionary epistemology and universal Darwinism directly because of the level of work required. It may be wiser to build off embodied artificial intelligence literature where researchers have provided a foundation of similar ideas.

It is encouraging to encounter Plotkin, 1993 and the fields of evolutionary epistemology and universal Darwinism because they lend weight to the notion of attempting an embodied artificial intelligence approach to learning and culture and because such work can contribute back to the two fields.

No comments:

Post a Comment