Object-classification as a decision tree  The Objective and the Biological Modes of Construal 
Object Perception
The faculty of segmenting the world into objects
(revised September 21, 1997) 

One of the main papers on this faculty is Elizabeth Spelke's "Principles of Object Perception" (1990); the work of Baillargeon (1987) on object permanence is a necessary supplement. Spelke and her associates have done pioneering work in experimental psychology, investigating the rich cognitive abilities of infants. In this article, she situates her work by contrasting it with the alternative traditions of Gestalt psychology and empiricism; in contrast to their proposals for domain-general cognitive faculties, she finds that children make specific kinds of assumptions about the nature of objects that cannot simply have been learned from experience (the frame problem precludes the possibility of bootstrapping oneself up from a position of no prior assumptions), nor are they the intuitively plausible ones of "figural goodness".

Spelke does not address the issue of how children come to have the cognitive dispositions she so meticulously detects and demonstrates. In that sense, her argument is clearly Kantian: early categorization depends on transcendental principles, in the sense that they are not derived from experience. Darwin's painstakingly developed theory of the gradual construction of organic form through natural variation and selection provides a way of thinking about the development of children's strikingly complex inferential abilities that was not available to Kant (cf. Lorenz 1977). This allows us to correlate the specific hypotheses children reliably make about a world that must otherwise have been wholly mysterious to them with the characteristics of the environment in which human beings evolved--the hundreds of thousands of years prior to the advent of agriculture and civilization--and the specific kinds of challenges human infants would have been faced with.

Spelke's Basic Experimental Design

Theoretical Considerations Differences with Other Theories Main Claim Experiments The Thesis of Centrality Some Entailments Spelke argues that this indicates "a link between processes of perceiving objects and processes of reasoning about the physical world" (51).

The criteria utilized to identify objects permit a range of inferences about the likely behavior of objects, and studies of children indicate that they infer that

The Use of Categories

Spelke's work forms the foundation for an understanding of categorization, discussed extensively by Markman (1990): once a category is formed, it becomes a potent site for storing information about members of that category, which can then be applied to new objects of the same kind. Spelke proposes that the identification of a number of items as objects permits the gathering of new general features about objects, beyond what was utilized to specify the objects in the first place. Such generalizations may give rise to the Gestalt principles of figural goodness, color, etc., thus explaining the adult dependence on such cues.

Markman, along the lines of Rosch and Lakoff, finds evidence that categories are organized in terms of family resemblances and prototypical category members. Spelke's findings suggests that some core features (and thus the typical family member) may be specified by nativist cognitive faculties, with an interesting consequence: the information stored in a category--the conceptual primitives, image schemas, and inference algorithms--may apply with the greatest accuracy to the prototypical members of that category. The properties of objects, for instance, do not apply with equal felicitousness to fluids and solids. While a rock is a prototypical example, wind and water are marginal, even doubtful members of the class of objects, and the core properties of objects fit them poorly.


The significance of Spelke's findings for Cognitive Culture Theory comes in at least a couple of flavors:

  1. They provide evidence for the existence of a rich and specified cognitive structure whose functional assumptions are not derived from experience. This implies
  1. They demonstrate the importance of enabling assumptions in the construction of the known world, thus suggesting
The first point, that of establishing the presence of cognitive structures for object perception that are not empirically derived, confirms a state of affairs denied by Locke, suspected by Hume, and argued forcibly by Kant; the debate resurfaces more recently in the debate between Skinner and Chomsky (cf. the Introduction). In its modern incarnation as the frame problem, the debate resourfaces with a new twist in The Sociobiological Fallacy.

The specific properties that infants assume objects have--cohesion, boundedness, rigidity, and the absence of action at a distince--are so fundamental that we easily loose sight of how complex they are. It was not until people attempted to construct machines that duplicate human behavioral and cognitive skills that the segmentation of visual arrays into stable three-dimensional objects began to be appreciated for the rich faculty it really is. A robot with arms and general instructions for motion, unless instructed in excruciating detail not to, will as a matter of course attempt to move its limbs through objects--including itself, leading to bizarre behavior such as trying to push an arm through its own head or body. Children, on the other hand, start out assuming that objects cannot occupy the same space and utilize this assumption to detect objects, accepting input from both touch and vision.

The second point takes us back to Galileo's distinction between "primary" (or objectively real) and "secondary" (or subjectively projected) qualities, and thus to the proclivity towards object-construals in science.

Philosophical materialism

In Galileo's materialist ontology, the crucial distinction is between "primary" and "secondary" qualities:

Spelke's analysis of the qualities children utilize to parse the world into objects takes less for granted, but Galileo's primary qualities remain within the orbit of the child's enabling concepts of cohesion (that all parts of an object move on connected paths over space and time), boundedness (that two objects cannot occupy the same place at the same time), and rigidity (that objects do not deform as they move). The fourth enabling assumption, that objects only affect each other if they make contact, formulates a major principle of object mechanics--a principle that led Galileo to reject the notion of gravity as "occult". Assumptions that have enabled thousands of generations to parse the world into objects are not necessarily useful when a greater degree of precision in knowledge is sought for; in fact, they may prevent us from making progress.

The thoroughly tested nature of the object-construal faculty may mislead us into equating its output with the world itself. We have a variety of reasons to believe that objects constitute a mode of construal that is both phylogenetically and ontogenetically older than the structures that enable us to perceive the world in terms of purposes and mental states. The most obvious is that cats and dogs reliably perceive the world in terms of objects--animals that have diverged from the hominid and primate lines millions of years ago have the same basic skills, indicating that our common ancestor also had them. In addition, object perception is a precondition for other kinds of cognition, such as the detection of agents and minds, as proposed by the decision tree in the previous page. The antiquity of the object perception faculty has given it a very long period of time for natural selection to act upon it and to perfect it by degrees. It is therefore fully to be expected that such a well-designed faculty would occasion a great deal of confidence in the world it construes, which goes some way towards explaining the intuitive appeal of the notion that the world is literally composed of cohesive, bounded, rigid entities that cannot affect each other at a distance. But even if we find it reasonable or convenient to believe that the world really is composed of such entities, on what grounds could we possibly hold that this is the only legitimate way of construing the world?

These modes of construal are additionally useful--and similarly perilous--because they can take input from a variety of sources, and thus permit their inferences to be translated into other domains. When objects are utilized as a source domain for the mapping of a cultural phenomenon, as takes place in the materialist psychological and social theories of the Enlightenment, the underlying assumptions of the object-construal faculty are responsible for much of the conceptual payout.

Quote of the day

"We fancy, that were we brought on a sudden into this world, we could at first have inferred that one billiard-ball would communicate motion to another upon impulse; and that we needed not to have waited for the event, in order to pronounce with certainty concerning it. Such is the influence of custom, that, where it is strongest, it not only covers our natural ignorance, but even conceals itself, and seems not to take place, merely because it is found in the highest degree." Hume 1748, Section IV, Part 1, paragraph 24.

Object-classification as a decision tree 
Next: Objective and Biological Modes of Construal
The Objective and the Biological Modes of Construal 
 Return to the CogSci index
© 1997 Francis F. Steen, Communication Studies, University of California, Los Angeles