Lectures on Perception

An Ecological Perspective

Michael T. Turvey

• Publisher: Routledge
• Year: 2018
• ISBN:
  9780429813382
• Edition: 1
• Topic: Psychologie
• Subtopic: Histoire et théorie en psychologie

Lectures on Perception: An Ecological Perspective addresses the generic principles by which each and every kind of life form—from single celled organisms (e.g., difflugia) to multi-celled organisms (e.g., primates)—perceives the circumstances of their living so that they can behave adaptively. It focuses on the fundamental ability that relates each and every organism to its surroundings, namely, the ability to perceive things in the sense of how to get about among them and what to do, or not to do, with them. The book's core thesis breaks from the conventional interpretation of perception as a form of abduction based on innate hypotheses and acquired knowledge, and from the historical scientific focus on the perceptual abilities of animals, most especially those abilities ascribed to humankind. Specifically, it advances the thesis of perception as a matter of laws and principles at nature's ecological scale, and gives equal theoretical consideration to the perceptual achievements of all of the classically defined 'kingdoms' of organisms—Archaea, Bacteria, Protoctista, Fungi, Plantae, and Animalia.

Part 1

Foundational Concepts

Lecture 1

What Kinds of Systems Do We Study?

In abstract terms, a beginning student of perception (and by implication, action) has embarked on the study of epistemic, intentional systems. The focus is in respect to the function of perception more than its content. Perception’s universal function is to coordinate the individual organism (specifically, its activity) with its surroundings. The content of perception, the specific varieties of an organism’s awareness of body and environment (see Gibson, 1979/1986), will vary with the organism-niche specifics. In this initial lecture, we begin with the concept of system, proceed to the qualifier intentional, and end with the qualifier epistemic.

1.1 System

The word system, like many of the important words in the study of perception and action, is vague. It refers to a loose concept, one that is readily intuited but not easily codified. This is not a damaging criticism. Science is infested with loose concepts that play constructive and creative roles (Elkana, 1970; Löwy, 1992). Denying them, or awaiting their strict codification before use, would hamper progress. That said, the needed deployment of loose concepts must be coordinate with the nurturing of respect for precise well-formulated concepts.

Given the potential for vagueness in defining system (e.g., Berlinski, 1976; Marchal, 1975), it might be more fruitful to begin with the opposite notion. What can be meant by non-system? A set of isolated pieces that don’t interact, or interact so weakly that their influences upon each other are negligible, seems to fit the bill. Even better, perhaps, is the notion of a collection of related pieces where the relations have no implications for the properties or behaviors of the pieces. Certainly lacking in the image of a non-system is the sense of shared influences or mutual dependencies; intuitively, a non-system exhibits no coherence or functional unity. Also lacking is the sense of a boundary, a separation of the pieces into “ground” (pieces that surround) and “figure” (pieces that are surrounded).

Heaps and aggregates are sometimes promoted as intuitive examples of non-systems (Bunge, 1977; Grobstein, 1973). In a heap or aggregate, the properties of any one component when it is excluded from the aggregate is the same as when it is included in the aggregate. A rock in a rock pile is related to the other rocks in terms of distance, weight, and shape, but these relations make no difference to the individual rock. Remove it from the pile and it possesses the same properties that it had in the pile. Minimally, one would think, a system is distinguished from a non-system by the fact that, in a system, the relation “makes some difference to its relata (Bunge, 1979, p. 6).” It helps to distinguish between what might be called bonding and non-bonding relations (Mahner & Bunge, 1997). A relation between two things x and y is bonding if and only if the states of y alter when the relation to x holds. The examples above of relations of distance, size and weight are non-bonding. A non-bonding relation such as the distance between x and y does not itself bring about a change in the state of y but it may allow that x can act upon y.

A problem with the use of heaps and aggregates as non-systems is that straightforward heaps (in the form of sand piles and rice piles, for example) and common biological aggregates (a collection of amoeba, for example) can, under certain conditions, exhibit self-organized coherent behavior (Bak, 1996; Garfinkel, 1987). Apparently, non-system is not a permanent designation. In recognition of this fact, the terms facultative and obligate have been suggested to distinguish, respectively, between (a) systems that assemble from and disassemble to non-systems and (b) systems that persist as systems once assembled (Grobstein, 1973; Juarrero, 1999).

Although our willingness to ponder the curious notion of non-system has proven fruitful, the enterprise fails to convey fully what system must mean because, for non-system, boundary is vacuous. A system, unlike a non-system, has a distinguishable “inside” and “outside.” A wall (or boundary, or interface, or dynamical process) separates and shields the inner from the outer components (Krieger, 1992). Of particular significance is how to interpret the outside. Is a system individuated by its inner components alone or by the inner and outer components together? The latter of the two alternatives is the answer given by Figure 1.1.¹

In Figure 1.1, we arrive at a definition or model of a system s by considering (i) the inner degrees of freedom (DF) that compose it, C(s) (ii) the outer DF with which it interacts, E(s) and (iii) the relations, both bonding and non-bonding, among inner DF and between inner and outer DF constituting its structure, S(s). This model comprising composition, environment and structure, a CES model, is a minimal starting point for understanding any system of interest at any level of interest (Mahner & Bunge, 1997). The environment E(s) within the CES model is always relative to a given system s. As such, an environment is not defined without a system (there are no empty environments), there are as many environments as there are systems (excluding the universe as a whole), and an environment of a system is not an entity, not itself a system.² Because environment is not an entity, it cannot be that E(s) as a whole acts upon or interacts with s; rather, members of E(s) act upon or interact with members of C(s).

Figure 1.1 permits an overview of scientific predilections. Inquiry restricted to C(s) defines reductionism. Inquiry restricted to E(s) defines environ-mentalism. Inquiry restricted to S(s) defines structuralism. Inquiry that gives due consideration to all members of the triple 〈C(s), E(s), S(s)〉 defines systemism (Bunge, 1979). Less dryly, we can contrast the latter with holism, the view that every thing is connected to every other thing, and atomism, the view that every thing operates in isolation from every other thing. For systemism, every thing is connected with some other thing or things (Mahner & Bunge, 1997).

Figure 1.1 The minimal model of any arbitrary material system is: m(s) = 〈C(s), E(s), S(s)〉. C(s) is the collection of all the parts of s, its composition. E(s) is the collection of all those things other than s that can influence s or be influenced by s, the composition of s’s environment. The thin arrows are the relations among the parts of s, its internal structure S_int(s). The thick arrows are the relations among s’s components and environmental things, its external structure S_ext(s). The union of the internal and external structure is S(s).

(From Figure 25, Bunge, 1979, adapted with permission from Dover Publications.)

1.1.1 Partial Systems

Taking nature apart to find systems is a necessary scientific strategy. It’s how we make nature manageable. It is, nonetheless, a strategy fraught with difficulty. There is always the risk of slipping by the system that is actually exhibiting the phenomena of interest. There are two unwelcome consequences of over-decomposing (Turvey & Shaw, 1979). First, the phenomena may appear indeterminate, lacking reference to any underlying law, when in fact, at their proper and coarser grain size of analysis, the phenomena are deterministic and lawful (Humphrey, 1933). Second, erroneous content or function may be ascribed to the partial system (Ashby, 1963; Casti, 1989). Where a selected system is at the wrong grain size of analysis for the phenomena of interest, that system must be endowed with properties that are conjectured to have the ability to generate the phenomena.

With respect to the first consequence, imagine that the phenomena of interest are (a) the short-period diurnal and semidiurnal tides and (b) the long period tides whose rhythms range from 14 days to 19 years. If attributed to the Earth-Water system, these short- and long-period oscillations in sea level would look capricious. Alternatively, they might invite the hypothesis that there are two different systems in operation on different time scales. In contrast, if these temporal tidal events are recognized as the phenomena of a larger system, the Sun–Moon–Earth–Water system to be exact, then they will be seen as lawful. The latter, more inclusive system embodies the laws of which the tides are necessary consequences. Those laws are not embodied by the Earth–Water system (Figure 1.2).

With respect to the second consequence, we can take an example from Ashby (1963). This example makes the general point that if the total system (from the perspective of the phenomena of interest) is unobservable, then the partial system that can be observed may assume “remarkable, even miraculous properties” (Ashby, 1963, p. 114). The paradigmatic case is the magician’s trick. It looks miraculous because not all of the significant variables are observable.

Consider a system composed of two interconnected devices A and B and the input I that influences both of them (Figure 1.3). Thus, A’s inputs are both B and I. The device A shows some characteristic behavior R only when B is at state z and I is at state y. It is the case that B is in state z only subsequent to I taking the value x. There are two observers. Observer 1 sees the total system and is able to conclude that R occurs whenever the total system shows a state with B at z and I at y. Observer 2 cannot see B (or does not take it into account). Consequently, knowing the states of A and I is insufficient to predict reliably the occurrence of R. After all, I is sometimes y and sometimes some other state. Nevertheless, Observer 2, by paying attention to earlier states of I can make reliable predictions about R. If I passes successively through states x and y, then R will occur and not otherwise. It follows, therefore, that Observer 2 can make reliable predictions by using successive values of I that are in fact observable.

Figure 1.2 Partial system (left) and total system (right) in the explanation of tides.

Figure 1.3 Two observers try to explain the behavior R of a system. Observer 1 sees the total system A + B + I; Observer 2 sees only the partial system A + I. Observer 1 explains R through the present states of the system; Observer 2 has to appeal to past states, to memory, to explain R. See text for details.

Suppose the two observers now get into an argument about the “system.” Observer 1...