Neil W. Rickert (1998) Knowledge is not Unitary. Psycoloquy: 9(50) Efference Knowledge (2)

Volume: 9 (next, prev) Issue: 50 (next, prev) Article: 2 (next prev first) Alternate versions: ASCII Summary
Topic:
Article:
PSYCOLOQUY (ISSN 1055-0143) is sponsored by the American Psychological Association (APA).
Psycoloquy 9(50): Knowledge is not Unitary

KNOWLEDGE IS NOT UNITARY
Commentary on Jarvilehto on Efference-Knowledge

Neil W. Rickert
Department of Computer Science
Northern Illinois University
DeKalb, IL 60115
http://ux.cs.niu.edu/~rickert/

rickert@cs.niu.edu

Abstract

In his attempt to account for the complexity of interactions between an organism and its environment, Jarvilehto (1998) has given us a theory in which organism and environment are considered to jointly constitute a single unitary system. While I support the idea of giving a larger role to the environment than is typical in theories of cognition, I will suggest that Jarvilehto goes too far, erasing the distinctions that are prerequisite to the very possibility of having a theory of cognition.

Keywords

afference, artificial life, efference, epistemology, evolution, Gibson, knowledge, motor theory, movement, perception, receptors, robotics, sensation, sensorimotor systems, situatedness
1. There has been a long tradition of treating the human as a rational agent, or at least a minimally rational one (Cherniak, 1986). This leads to the ideas of folk psychology, in which a cognitive agent attempts to achieve desires by applying logical methods to observations and beliefs. However, rational agency models have failed to adequately account for human behavior, as has been demonstrated by research such as that of Kahneman, Slovic and Tversky (1982).

2. Rational agency models depict an organism as an input-output system. The system receives input signals from its sensory organs and carries out some type of quasi-computational process which makes use of both sensory inputs and the accumulated knowledge (beliefs, memories, represented data) in order to direct the production of outputs (motor actions, speech). While rational agent models of cognition are still viewed positively by many in psychology, philosophy, cognitive science and artificial intelligence, there is a small group of dissenters. Among them we find Maturana & Varela (1987), Winograd & Flores (1987) and Rose (1998). In his target article, Timo Jarvilehto (1998) has aligned himself with this group.

I. KNOWLEDGE

3. The general criticism of rational agency models has been that by isolating the input from the output, the problem for an agent is greatly oversimplified and that this ignores the complexity of the interactions between an organism and its environment. We might say that some form of interactionism is the suggested alternative to rational agency.

4. In his theory, Jarvilehto introduces an unusual way of describing this interaction. His idea is that the organism and its environment should be considered together as a single unitary system (paragraph 4). Jarvilehto uses the formation of knowledge as a vehicle to show how his approach can be used.

5. Knowledge has always been seen as related to interaction with the environment. The rational agent interacts by applying rules to the inputs, so as to determine the outputs. It has seemed natural that the rules applied should be considered knowledge, and this leads directly to the idea of knowledge as a system of beliefs, or a system of true beliefs that the agent justifiably holds. Traditionally, the inputs received from perception have also been considered knowledge (perceptual knowledge); this has given us the idea that knowledge formation is some kind of input process, where in some sense the data that have been put in are transferred into knowledge in the form of memories.

6. Jarvilehto rejects the idea of knowledge formation as a transfer process, and with it he rejects the idea of knowledge as justified true belief. In rejecting this idea of knowledge, he joins other proponents of interactionism. That it must be rejected is obvious from Jarvilehto's unitary model. As he says, "knowledge formation cannot be based on any transfer process from the environment into the organism, because there are not two systems between which this transfer could occur." This approach provides an easy way in which Jarvilehto can reject traditional accounts of knowledge acquisition. Regrettably, he has failed to provide us with a satisfactory alternative.

7. Here is what Jarvilehto says about the process of knowledge formation:

    New knowledge is created by perception when new parts of the
    environment join the organism-environment system while changing its
    structure. An increase in knowledge means a widening and
    differentiation of the system, which makes new kinds of behavioral
    acts and new results of behavior possible. [para. 4]

The difficulty is, that if we must consider knowledge in terms of a single unitary system, then there is no entity to which we can attribute the knowledge, other than that single unitary system.

8. If the fox learns some new tricks which enable it to better catch rabbits, we would normally want to say that this was new knowledge for the fox. And if the rabbits in turn learn to elude the fox and thus evade those new tricks, we would want to consider this as new knowledge for the rabbits. But if we must consider a unitary system, we can only say that with the new tricks the overall system has changed its knowledge as there was new system behavior. But as the rabbits found how to evade the new trick, the overall balance of behavior dropped back to where it was, and the new system knowledge was lost. By joining the entities into a single unitary system, we have removed the distinctions required if we are to ascribe knowledge.

9. What I have described is clearly not what Jarvilehto intended. Throughout the target article he talks of the organism as the entity which has knowledge. Thus he talks of the organism knowing something which allows it to act in its world (paragraph 39). The curious thought experiment (paragraphs 24-35) is all about the organism obtaining new abilities. Presumably Jarvilehto is not taking the single unitary system quite as seriously as he suggests. He owes us a better explanation of how the idea is to be used; and he owes us a clearer statement of what he means by "knowledge."

II. MOTOR ACTIONS

10. To Jarvilehto, motor outputs and efferent neural signals are important aspects of an organism's interactions. He contrasts this with traditional ideas based on the theory of information, and on rational agency and cognitivist ideas about cognition. The target article is at its best in this type of comparison. More traditional approaches have tended to treat an organism as if it were a passive receiver of information and have proposed methods such as induction by which the organism can form knowledge from that input information. By contrast, Jarvilehto suggests a far more dynamic organization of interactions between input and output.

11. The idea that output activity is an important part of input activity is hardly new. It is an idea that J.J. Gibson stressed in his theory of direct perception (1966, 1972, 1979). Gibson emphasized the extent to which the person is an active participant in perception. Jarvilehto (paragraph 2) seems to criticize Gibson for considering the organism and environment as two separate systems resonating with one another rather than as a single unitary system. Yet, as I have argued above, Jarvilehto must himself distinguish between organism and environment if he is to have a viable concept of knowledge.

12. William Powers (1973), with his control theoretic ideas about cognition, is another researcher who has given us a picture of a dynamic system of inputs and outputs connected in a complex feedback system. His ideas seem to be widely ignored today, and that is a pity. Perhaps Jarvilehto will be able to find something of value in them.

III. THE ROLE OF INPUT

13. What I find strangest about the target article is the position that Jarvilehto takes on input. In his zeal to demonstrate that knowledge is not something which is transmitted through the sensory input system, he seems to argue that inputs are not all that important. In a rather curious thought experiment (paragraphs 24-35), he purports to show that an organism can acquire knowledge without any input at all. Yet Jarvilehto is surely mistaken about this. As he himself admits (paragraph 31), the organism does receive inputs by virtue of receiving energy from its environment.

14. The thought experiment fails to demonstrate that knowledge can be acquired without input. If anything, it demonstrates the need for input, in that without the energy inputs the organism of the thought experiment could have done nothing to improve its performance. However, the experiment is more concerned with arguing that a system of fixed receptors will not do. The latter part of the experiment shows how an organism can effectively use its abilities to adjust its sensory inputs. This is consistent with the view that perception is an active process (Gibson 1966), and that feedback is important (Powers 1973).

15. Jarvilehto is specifically concerned with refuting a particular idea about how input is used in cognition. He is objecting to the assumptions made in rational agency theories of cognition. He seems to be unaware of other possibilities, such as the control theory of Powers, or the perceptual learning theories of Eleanor Gibson (1969). There is an irony here. Jarvilehto is primarily concerned with the complex interactions between an organism and its environment. Yet, in downplaying the role of inputs, he is arguing against one of the core requirements for such complex interactions.

16. If the main aim of the target article is to refute the idea that knowledge is transmitted through the sensory systems, Jarvilehto would do better to examine other possible ways in which sensory inputs could be used for forming knowledge. Piaget (1954) has suggested that the child's process of forming knowledge is one of construction, carried out in a program that he calls genetic epistemology. Eleanor Gibson (1969) has described a method of forming knowledge which she calls perceptual learning. In perceptual learning, the organism improves its ability to make input discriminations. Indeed, the type of knowledge formation that Jarvilehto describes in his thought experiment would seem to be an excellent example of perceptual learning.

IV. SUMMARY

18. Professor Jarvilehto has given us an intriguing way of looking at the relationship between an organism and its environment. Regrettably, in the form in which he has presented it, his unitary approach would seem to deny us the conceptual distinctions that are necessary to talk about knowledge and its acquisition. Yet I think it likely that with suitable adjustments he could refine his model into one that is better able to address the problems that interest him.

REFERENCES

Cherniak, C. (1986). Minimal Rationality. The MIT Press.

Gibson, E. J. (1969). Principles of Perceptual Learning and Development. Appleton Century Crofts.

Gibson, J. J. (1966). The Senses Considered as Perceptual Systems. Houghton Mifflin.

Gibson, J. J. (1972). A theory of direct visual perception. In Royce, J. R. and Rozeboom, W.~W., editors, The Psychology of Knowing, pages 215-227. Gordon and Breach.

Gibson, J. J. (1979). The Ecological Approach to Visual Perception. Houghton Mifflin.

Jarvilehto, T. (1998). Efferent influences on receptors in knowledge formation. Psycoloquy 9(41) http://www.cogsci.soton.ac.uk/cgi/psyc/newpsy?9.41 ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1998.volume.9/psyc.98.9.41.efference-knowledge.1.jarvilehto

Kahneman, D., Slovic, P., and Tversky, A., editors (1982). Judgment under Uncertainty: Heuristics and Biases. Cambridge University Press.

Maturana, H. R. and Varela, F. J. (1987). The Tree of Knowledge: the Biological Roots of Human Understanding. Shambhala, Boston.

Piaget, J. (1954). The Construction of Reality in the Child. Basic Books, Inc.

Powers, W. T. (1973). Behavior: the control of perception. Aldine Publishing Co.

Rose, S. (1998). Lifelines: Biology Beyond Determinism. Oxford University Press. ftp://ftp.princeton.edu/pub/harnad/BBS/.WWW/bbs.rose.html

Winograd, T. and Flores, F. (1987). Understanding Computers and Cognition. Addison Wesley.


Volume: 9 (next, prev) Issue: 50 (next, prev) Article: 2 (next prev first) Alternate versions: ASCII Summary
Topic:
Article: