Monday, September 19, 2005

Factors influencing the origins of colour categories 

Factors influencing the origins of colour categories
author: Tony Belpaeme
"In the evolution of language field a large body of work exists on mapping meaning to form, but little attention has been paid to the formation of meaning itself. Nevertheless, the debate on the nature of perceptually grounded concepts such as colour categories, and the impact of language on this, is central to cognitive science and linguistics. One point of view considers perceptual concepts as universal to all humans and argues that genetical determinism is responsible for this. Opponents adhere to a relativist account, which claims that concepts are learned and thus are ecologically and culturally specific.
Colour categorisation presents an opportunity for testing both perspectives. Berlin and Kay, in the late sixties, and Rosch, in the early seventies, provided convincing evidence for the universalist position through linguistic and memory experiments. Field data collected and interpreted over the last decades show a remarkable agreement between colour categories from different cultures. The relativist position, it is argued, is unable to account for these phenomena. Backed up by results from neuropsychology on the opponent character of colour perception the universalist stance has held strongly. Only recently, evidence from anthropology and close scrutiny of experimental results and the deductions based on these undermine the authority of the universalist model. The question arises if the relativist position can account for the strong facts brought forward by experimental data. This work investigates outstanding issues in the colour categorisation debate using computational modelling. It reports new insights on the plausibility of accounts of the origins of colour categories. Four simulations have been constructed, two in which colour categories are evolved through a process of natural selection, the other two in which colour categories are formed with a learning approach using environmental, ecological and cultural constraints. Both genetic evolution and learning are studied with and without the influence of language.
The basic entity of the simulations is the agent. The colour perception of an agent is modelled as a mapping from spectral measurements to an internal three-dimensional colour space. Categories are defined on this internal space as adaptive network. These adaptive networks consist of locally tuned receptors sensitive to small region in the space. The output of a network is the weighted sum of the reactions of all its locally tuned receptors. Categories can also be associated with words, needed for communicating colour meaning to other agents. The association between a category and a word is modelled by a scalar value representing the strength of the association."

From Multidimensional Signals to the Generation of Responses 

From Multidimensional Signals to the Generation of Responses
author: Gosta H Granlund
"Abstract. It has become increasingly apparent that perception cannot be treated in isolation from the response generation, firstly because a very high degree of integration is required between different levels of percepts and corresponding response primitives. Secondly, it turns out that the response to be produced at a given instance is as much dependent upon the state of the system, as the percepts impinging upon the system. The state of the system is in consequence the combination of the responses produced and the percepts associated with these responses. Thirdly, it has become apparent that many classical aspects of perception, such as
geometry, probably do not belong to the percept domain of a Vision system, but to the response domain.
There are not yet solutions available to all of these problems. In consequence, this overview will focus on what are considered crucial problems for the future, rather than on the solutions available today. It will discuss hierarchical architectures for combination of percept and response primitives, and the concept of combined percept-response invariances as important structural elements for Vision. It will be maintained that learning is essential to obtain the necessary flexibility and adaptivity. In consequence, it will be argued that invariances for the purpose of
vision are not geometrical but derived from the percept-response interaction with the environment. The issue of information representation becomes extremely important in distributed structures of the types foreseen, where uncertainty of information has to be stated for update of models and associated data."

"invariant representation" "conceptual structure" - Google Search 

"invariant representation" "conceptual structure" - Google Search

Programming Considerations for a Brain-Like Computer 

Programming Considerations for a Brain-Like Computer
"We feel computers with a brain-like architecture will be required to run efficiently cognitive applications now under development. Examples of such applications include natural language processing, text processing, intelligent data mining, Internet search, human-computer interface, decision-making, and artificial intuition.
A brain-like computer is a fundamentally different device than a traditional digital computer in both hardware and software. Programming one will require different ways of thinking about computation along with extensions of current hardware and programming languages. They will do well in many cognitive applications where current computers do poorly but they are also likely to do poorly at the tasks where current computers excel. We conjecture a brain-like computer will act like an unfamiliar combination of an analog computer and a manipulator of discrete entities. Some of the unusual operations used by the brain-like programs we have developed as part of the Ersatz Brain roject are:
1. Programming with the topography of the data representations.
2. Topographically determined formation of feature combinations based on lateral spread of patterns.
3. Control of the computation by use of topographic representations: weighting using “programming patterns”.
4. Use of distributed data representations involving the array as a whole.
5. Use of a dynamical system for (a) the integration of multiple inputs and (b) the selection of a single output response.
6. Similarity computations between very large state vectors."

"Memory-prediction framework" - Google Search 

"Memory-prediction framework" - Google Search

Memory-prediction framework 

Memory-prediction framework - Wikipedia, the free encyclopedia
"The memory-prediction framework theory of the brain, created by Jeff Hawkins and described in his book On Intelligence, argues that the brain works mainly by the neocortex's ability to match sensory input to stored patterns and predict what will happen next.

The theory is motivated by the observed similarities between the brain structures (especially neocortical tissue) which is used for a wide range of behaviour available to mammals. The theory posits that this remarkably uniform physical arrangement of neural tissue reflects a single principle of complexity management which underlies all cortical information processing. The basic processing principle is hypothesized to be a feedback/recall loop which involves both cortical and extra-cortical participation (the latter from the thalamus and the hippocampus in particular).

The memory-prediction framework provides a unified basis for thinking about neurobiology. Although certain brain structures are identified as participants in the core 'algorithm' of prediction-from-memory, these details are less important than the set of principles that are proposed as basis for all high-level cognitive processing."

This page is powered by Blogger. Isn't yours?