Final draft copy
Playing Against the Self: Representation and Evolution
David Myers
Loyola University
Abstract

This paper discusses the evolutionary role of play in creating and extending representational form. Specifically, the paper examines the egocentric nature of play from a natural-historical perspective. Given the widespread tendency toward egocentric behavior during play, what is the function of this behavior in the evolution of a playful species such as our own? This question is answered as regards the benefits of species variation wherein play is considered a stochastic-like mechanism inducing cognitive variations analogous to other mechanisms (eg., sexual reproduction) inducing physical variations. The discussion considers theoretical support and implications of a playful evolution.
 

Playing Against the Self: Representation and Evolution
Body (draft copy)

I would like to discuss how important it is that we play "against" something when we play. Now, playing "against" something seems to be more obviously the case when we play games of various sorts than when we engage in, for instance, free play or daydreaming or the like. But I would like to develop the argument that it's not and that, in fact, playing "against" something is what happens all the time, whenever we play.

Admittedly, this requires defining play somewhat apart from all behavioral manifestations of the play phenomenon. Herein, I will conceptualize play as a transformational process operating most fundamentally on mental representations.[1] Thus, physical mimicry -- and the like - would be considered play only to the extent that physical mimicry would be associated with mental (symbol) transformations. I will adopt this conceptualization fully realizing that playful behavior seems to exist prior to our species' representational competencies and should have, for that reason, physical as well as mental consequences among a variety of organisms. Nevertheless, I will consider the mental consequences of play of a different sort and, ultimately, of more significance than its physical consequences.

In any case, what I would like to make clear from the beginning is that I do not by this argument mean to imply that play is of two sorts: the "against" sort and the other sort. I would wish to maintain that it's all the same sort, regardless of what consequences are involved. It's all "against."

And, if so, then against what?

A model of representational evolution.

The accumulation of characteristic variations leading to speciation is, within Darwinian evolution, a delimited game. The fittest survive and the winners win -- within some bounded ecological niche -- only to the extent that the non-fit lose and the losers die.

The great distinction of play -- perhaps its most significant distinction -- is that the losers of play do not die; they run away (or, more normally, simply submit) to play another day. This speculation -- that a more playful evolution might well transform Darwinian notions -- is the basis of this essay.[2]  It leads us to consider how a representational species might evolve beyond the context of its natural-historical environment. Here is the question I then pursue: does the evolution of representational characteristics and skills (wherein I would define play -- see Myers, 1991) differ from the evolution of non-representational characteristics and skills in any fundamental way?[3]

For instance, it seems as though the evolution of representational characteristics and their associated skills has occurred much more rapidly than the evolution of non-representational characteristics and skills. Though information about these matters is imprecise, the development of homo sapiens sapiens from homo habilis over a period of (approximately) two million years involves a series of quite radical changes.
 

But, unfortunately, not only is the exact nature of these changes uncertain, the uniqueness of their rapidity is also controversial. There are several related controversies within evolutionary theory (see Mayer, 1991, for a list); the most pertinent here is that between "continuity" and "discontinuity" models of evolution. Either species characteristics evolve through a series of random and rather abrupt leaps (from australopithecus to habilis to erectus to sapiens, for instance), or species characteristics evolve more gradually through much smaller and more continuous variations.

The latter of these two possibilities is the continuity model, and, all in all, the more conservative and traditional of the two. If, indeed, evolution is fundamentally continuous, then hominid evolution, which appears very choppy and discontinuous, is unique and conspicuous in that choppiness and those discontinuities -- which we then might well attribute to the unique and conspicuous representational skills of the human species. If, however, evolution is fundamentally discontinuous (as championed by Gould, 1977, among others) then the rapid evolution of hominids is less unique than it would at first appear -- and could not so easily be attributed to representationalism alone.

Each of these two models of evolution -- continuity and discontinuity -- has its own set of beliefs and assumptions. And each of the models has its unsolved problems. The discontinuity version, for instance, must explain -- particularly as regards the seemingly over-rapid proliferation of the representational skills of the habilis line -- how such random, lucky, and improbable changes ever occurred, in succession, in such a limited time frame. One explanation is teleological -- it happened or we wouldn't be here to talk about it -- and, for that reason, less than satisfying. Another explanation is that rapid change is the result of the interaction between wide variations in species characteristics and a rapidly changing environment. However, it is unclear how such wide variations in species characteristics could be tolerated: would not species characteristics gravitate toward some mean value offering the relatively greatest survival potential for members of that species?

Pondering this last question leads down some interesting paths and requires extended discussion. At least the kernel of an answer to this question is in the quote below, if I can, for the moment, equate "structural change" with species' characteristics and "behavioral change" with species' skills:

This passage seems a tad misconceived in implication and, as such, indicative of a common misconception that misdirects thinking concerning cognitive evolutionary change, wherein some amount of structural change must always precede behavioral change.

If, for instance, in the first example in the quote above, the non-aquatic species were unable to forage in shallow coastal waters to begin with, it would either continue to live on land or die and become extinct. That is, there must be some "latent" characteristic(s) of that -- or any -- species that allows it to survive within an environment prior to its "adapting" to that environment. This is a subtle point, but an important one, because it means that the second example above is more the rule than the exception. Species adaptation -- and thus, over the long term, survival -- most often results from the proliferation of characteristics that are imperfectly adapted to the species' currently surrounding environment.[4]

Such characteristics would retain the species' capability to exist in alternative environments; most predictably, of course, these alternative environments would be those that had been amenable to that species prior to some previous sequence of differentiation -- due to the maintenance of vestigial forms. But these alternative environments could, in fact, co-exist with the species' current environment without being either superseded or masked by that environment.

The treecreepers above, for instance, discovered a new environment (of breakable tree barks) as a result of their imperfect adaptation (variations in beak size) to their old environment (of cracked, but impervious, tree barks) -- despite being under little or no ecological pressure (I would assume) to do so. In fact, the lack of ecological pressure (of which the prime example is competition) in this case would make "adaptation" more likely to occur (since the larger the population of treecreepers, the more likely a wide range in treecreeper beak sizes).[5]

However, a large variance of species characteristics would only be of benefit in inverse relationship to the disadvantages of the abnormalities produced. In the case of the treecreepers, for instance, an under-extended beak is most often simply a malformation, gaining its owner no significant advantage in bug-seeking and, in fact, very likely putting its owner at a disadvantage when competing for food with its sharper beaked brethren. And, in any short-lived competition between two treecreeper species identical in all respects but one -- the range and frequency of beak abnormalities -- the treecreeper species with the broader range of abnormalities would no doubt lose.

Yet, despite disadvantages incurred, it is constant variation in species characteristics of this kind -- random and common and seemingly apart from gross mutation -- that is the only reasonable explanation for the rapid evolution of organisms under sudden and severe ecological pressures.

The evolution of cadmium resistance [in aquatic relatives of the earthworm] could have taken no more than 30 years. In fact, [measurements] indicated that the degree of metal tolerance observed could have evolved in just two to four generations -- or a couple of years...

Either the genetic code of the "pests" above continually produces toxin-resistance variations (though few in number and with a low survival potential in non-toxin environments) or these various pest species must depend on very random, very lucky, and very improbable mutations each time a toxin is introduced into their environment. It is much more efficacious to assume the former than the latter.
***

Now, in the case of treecreepers and flatworms the relationship between characteristic and skill is rather straightforward: e.g., the characteristic of toxin-resistance entails the skill of surviving toxin-infested environments. But such relationships are much less clear regarding the cognitive characters and skills of homo sapiens.

Based on medical and neurophysiological evidence (Gazzaniga, 1992), it clearly seems that the sensory apparati of homo sapiens is wired for representationalism just as it is wired for sight, sound, and smell. But, in many quarters and regarding a great many cognitive issues, huge uncertainties remain concerning the location and precise nature of the imbedded representationalism characteristics of homo sapiens -- resulting in the inability to correlate these characteristics of representationalism with associated behavioral skills.

For instance, although we know the range of electromagnetic waves that the human eye is capable of perceiving, and we know the range of decibels that the human ear is capable of hearing, we do not know with any comparable degree of accuracy the range of representations the human mind is capable of representing. This is not from lack of trying. Attempts to determine this range have resulted in a great number of scales (Stanford-Binet and the like) that clearly establish ranges and variations of human cognitive skills. However, these sorts of measurements have tended to measure much more of skills and much less of characteristics than do similar scales of visual or aural acuity (resulting in, among other things, arguments over just what the characteristics of the "intelligence" skill actually are. (See Gardner, 1983, and Sternberg, 1988).

Further complicating our inability to precisely locate the neurophysiological mechanism(s) of representationalism is the realization that representational characteristics are not associated with skills -- as sight and hearing are -- that directly interact with our natural-historical environment. Therefore, the only immediate and direct measurement device that we have to indicate the degree of our representational "acuity" is representationalism itself. And, as post-structuralism makes clear (if anything), it is a major problem to measure anything with itself. This is a problem for evolution as well.

The single measurement of classical Darwinian evolution is a test of fitness, of survival. And this test is inevitably a test of competition, or comparison, between two or more skills within the same natural-historical environment. Whereas these natural-historical tests of fitness might apply as equally to non-representational characteristics as non-representational skills -- ie., toxin-resistance and toxin-survival -- these same tests are not as equally applicable to representational characteristics and representational skills. For instance, when comparing representational and non-representational characteristics and skills, one is immediately struck by the relatively vast over-complexity of representational skills. We can represent, it seems, an infinite number of totally false things. And, furthermore, the great majority of these false representations are wholly beyond any conceivable natural-historical test of fitness. They are, in effect, entirely meaningless to species survival.

Given this enormous amount of representational gobbledy-gook, it is fortunate indeed that representational skills have physical and thus natural-historical, evolutionary consequences only to the extent that these skills actually result in physical manifestations (eg., a behavioral change). Representational environments ("worlds of the mind") can then substitute for natural-historical environments as measurers/testers of species survival skills solely to the extent that these representational environments are enforced and maintained beyond some isolated and individual set of interior beliefs and conceptualizations (ie., representationalism itself).

But, here, then, is the rub: the complexity and variety of representational skills is made possible by the isolation of representational skills from direct interaction with the natural-historical environment -- leading to greater species adaptability. Yet simultaneously and for the very same reasons, our species' natural-historical environment cannot serve as a test of fitness for the vast majority of representational "skills."

Thus, in separating the world of the mind from the world of the senses, a species such as our own might well gain the ability to retain a great degree of variance in characteristics and skills, but to what end? If these variations are not directly exposed to environmental hardships -- ongoing tests of fitness --then there is no species-independent mechanism available to choose among them. And, to the extent there is no species-independent mechanism to choose among them, they will propagate equally and randomly throughout the species. Furthermore, to the extent that these capabilities are directly exposed to environmental hardships (this being the final and culminate part of the rub), they are ultimately indistinguishable from non-representational characteristics offering equivalent survival skills.[6]

Given the lack of an environment to serve as proper context for the variation and natural selection of representational skills, it is unlikely that such skills could have evolved. Yet they have evolved. I postulate here that such an environment is a part of the skills themselves. That is, representationalism might originate as a positive evolutionary adaptation only to the extent that its skills are varied among individual members and these skills are selected by the natural-historical environment as more or less fit. The only natural-historical environment directly available to select among representational skills is the indirect one of representationalism itself, or -- in the context of human beings possessing those skills -- the abstraction of self.[7]

According to this model, symbol-transformational play functions in dialectical opposition to itself: as adaptive mechanism as well as environment in which that adaptive mechanism must adapt. This perspective echoes other insights into the dialectical nature of play...

This theoretical approach, if extended beyond the context of representationalism, might also be placed among those more recent attempts to address what seem to be directional (and thus anti-Darwinian) trends in natural evolution. But along this path, I am not, at the moment, prepared to blaze; therefore, the argument will remain restricted to the evolution of representationalism and, specifically, representational play.[8][9]

The specific function I would set for play is this: to serve as a homeostatic mechanism increasing the variances of representationalism, not so much in genetic pre-determination as in phenotypic expression. So that, in the absence of natural-historical pressures to evolve, representationalism might well evolve nevertheless, in response to those ecological pressures created by its own presence within the species: ie. through play. This play mechanism would serve to increase combinatorially the formal processes of representationalism, but would not, in and of itself (at least not initially), assure that either survival benefits or survival negatives would be subsequently associated with these increasingly more combinatorially complex forms.

Thus, once the representational characteristics of the human brain were somehow minimally in place, play would serve as representational corollary to the random variations of physical sexual reproduction.

Observational and theoretical support for the model.

The common tendency for players to anthropomorphize the artificial intelligence of computer game opponents -- a tendency I would characterize as the egoism of play, and, as such, not restricted to the play of computer games -- seems rather odd until it is viewed within an evolutionary framework such as the one presented above. I could, for instance, refer to similar observations in a wide range of literature, perhaps exemplified by Piaget's observations of "egocentric" play.

Piaget's notion of egocentric play postulates an early state of cognition within which children are incapable of entertaining the notion that other minds exist in any form other than their own. That is, children under the ages of 7-8 are "permanently under the impression that they understand each other, and have no suspicion of the egocentric character of their thought." (Piaget, 1928, p. 207) This sort of egocentric thought might well occur among adults as well -- though Piaget believes that such adult egocentric thought "impoverishes analysis and consciousness of self" (Piaget, 1928, p. 209) without subsequent reflection and progression beyond its limitations. Even at more advanced cognitive stages, however, the distinguishing characteristic of egocentrism remains the prioritization of satisfaction of thought over objectification of thought. And, should we, as Piaget does, postulate such an early stage of cognition wherein egocentrism is as pleasurable as it is unavoidable, then it is at least reasonable to expect the characteristics of that particular and pleasurable cognitive stage to recur during the phenomenon of play.

There are innumerable examples of egocentrism during computer game play. One is the consensual denigration of computer games that "cheat" -- defined, in computer game player terms, as computer game opponents not having to abide by the same rules as their human counterparts. Thus, in the computer game Civilization, the game "cheats" insofar as playing pieces (boats, guns, etc.) controlled by the computer can move further and faster than those same pieces controlled by the human player. And, in the computer game Master of Orion, the game "cheats" insofar as computer opponents can utilize its playing pieces faster and at less cost than that incurred by the human player(s). These sorts of "cheats" are universally acknowledged as necessary to game play (to achieve proper balance between human and artificial intelligence), yet remain, simultaneously, a detriment to game play that would best and most enjoyably proceed without having to resort to such unfortunate kludges.
 

Similarly, whenever there is some doubt as to the actions and motives of computer game opponents (as occurs during "fog of war" or other hidden movement conditions), these actions and motives are initially assumed to be the actions and motives that would be most likely of human players in similar situations. Thus, there is the common reference to what the computer is "trying" to do, and how "dumb" or "smart" it is in the context of various game situations. These references occur among quite sophisticated computer game players, who are, undoubtedly, aware of the limitations of computer game opponents and, in fact, are quite capable of adopting game playing strategies that take full advantage of this or that non-human peculiarity in the game's artificial intelligence. Nevertheless, despite this knowledge of the computer game's inability to properly mimic human play behavior, it is apparently more pleasurable to play in ignorance than in possession of that knowledge. Those players who do find some way to beat or "cook" a computer game by adopting odd winning strategies -- that would obviously never achieve similar success against a human opponent -- point out such strategies as evidence of serious flaws in the game design. And, once discovered, these strategies make the game considerably less fun to play.
  This phenomenon of assigning qualities of the self to play mates/objects -- the egoism of play -- has been noticed elsewhere as well, but its significance has been uncertain. Dennett (1981), for instance, recognized early the common attribution of human motives and desires to non-human mechanisms (such as computer game opponents) -- but saw this as more fundamental to cognition in general than to play in specific. These attributions (eg., a chess-playing computer likes to "get its Queen out early" -- Dennett (1981)) seemed indicative of non-explicit mental representations -- that is, representations that have no explicit tokening in the artificial intelligence routines producing the behavior that is so represented. Yet it remains unclear as to exactly why these commonplace, non-explicit representations would be of exactly the sort one would attribute to a human being exhibiting such behavior. It is also unclear, within these philosophy-of-mind debates, why such attributions would have such a large impact on the pleasure experienced during computer game play.

A somewhat similar issue arising in an entirely different context concerns the Vygotskian notion of zones of proximal development or "zo-peds" as fundamental to the learning process. Within this approach to child development, learning is understood as a process of movement between an "actual development level" and "potential development as determined through problem solving under adult guidance or in collaboration with more capable peers." (Vygotsky, 1978, p. 86; italics added) The characteristics of these "peers," of course, will vary in accordance with the learning child. In all such cases, however, learning occurs most readily in response to relatively small variations in the familiar context of the child's self (the "zone of proximal development").

Further, the entertainment one derives from literature, when placed in an information-theory context, shows patterns of self-reference similar to those encountered in play-based models of learning. Paulson (1991), for example, describes a model of cognitive growth in which cognitive processes are continuously (recursively) reorganized during successive readings of a difficult poem or passage of prose.

In the Paulson model, the complexities of literature are first recognized only as "noise" (abnormalities) in the communication channel. Confronted with this noise, cognition (from our point of view, simply "play") contextualizes this noise as complexity within existing self knowledge. During this contextualization (symbol-transformational) process, self knowledge is reorganized and extended into previously unknown realms. This reorganization and extension, however, occurs in incremental micro-stages that imply continual, slight variations in self. And this reorganization and extension only takes place wherein there is some recognizable and confrontational (oppositional) difference between complexity and noise, or between self and not-self.

Theoretical implications of the model.

Why ever should any of this be so? That is, why should cognitive play (regardless of its conceptualization as learning or entertainment) most commonly and comfortably take place within the familiar context of self? The best sort of answer is, I would suggest, a natural-historical one describing the cognitive evolution of the species and involving, most particularly, the self-reflexive characteristics and skills of representationalism.

Ultimately, this answer would attempt to subsume the study of human physical play within the study of human mental play, and the study of human mental play within the study of cognitive evolution. It would represent a significantly different approach from most past investigations of play, since the study of evolution and the study of play have taken place most often in separate realms of theory and assumptions -- to the detriment, it seems, of both.

Play theorists, in particular, have seldom considered play strictly within an evolutionary context.[10]  Perhaps this is because many studies of human play have been anthropological in nature, and anthropological approaches often categorize play in accordance with the description found in Hunter's & Whiten's (1976) Encyclopedia of Anthropology: "The concept of play is never applied to behavior which meets individual or group survival needs." (p. 307)

There is also, of course, the widespread tendency to taxonomize play behaviors and then draw theoretical conclusions concerning play functions based on similarities among those behaviors -- as documented in Schwartzman (1978) and Chick (1989). This has led over time to the implicit and widespread assumption that play is most fundamentally a physical activity; thus we associate play more often with the characteristics of the child who is physically active than with the characteristics of the adult who is not.

These sorts of misconceptualizations regarding play seem to parallel those found in some aspects of learning theory, which has become synonymous in many contexts with a theory of child development. In contrast to Piagetian notions, recent research efforts (see Sternberg, 1990) indicate that learning strategies and intellectual development continue beyond the physical maturation process. Likewise, play theorists (Sutton-Smith, Bateson) have begun to question the boundaries and limitations established by past social-scientific theories of play -- particularly confining and defining play "as contributing to child development in one way or another." (Sutton-Smith & Kelly-Byrne, 1984, p. 308). In opposition to this trend, Csikszentmihalyi & Csikszentmihalyi (1988) have located play-like phenomena ("flow") within a variety of human experiences; and Bateson (1994) has been influential in particularizing play as a representational (framing) mechanism operating within an ecological context.

Despite the contrary of Bateson, evolutionists have also infrequently considered play as a critical factor influencing the development and maintenance of high-level cognition. While cognitive evolution remains a matter of great interest (and great mystery -- see Eccles, 1989) to many evolutionists, it is hard to find any reference to play per se in their work. There are, however, some interesting corollaries -- particularly in evolutionary biology. "Generativity", for instance, is very frequently mentioned as a unique and distinguishing characteristic of human cognition, particularly regarding high-level cognitive skills such as language. Corballis (1991) postulates a "GAD" (generative assembling device) somewhere within the left hemisphere of the human brain. This would, assumedly, be equivalent to a representational characteristic (or genotype) of what I have been calling "play."

Similarly, Edelman's (1989) theory of "neural Darwinism" assumes a high degree of human brain plasticity that would allow generative assembling devices (of whatever various sorts) to have a significant effect on brain structure and function.

And, paralleling the notion developed in this essay that the rapid cognitive evolution of the recent hominid line is the result of the playful combination of a finite number of pre-existing representational characteristics of the human brain, Gazzaniga (1992) and other selection theorists argue that evolution as a whole results from "the organism searching its [preexisting] library of circuits and accompanying strategies for ones that will best allow it to respond to the challenge" (p. 4). Likewise, Trehub (1991) proposes "an essentially deterministic cognitive model" (p. 297) of the brain, yet recognizes the importance of invention as "arguably the most consequential characteristic that distinguishes humans from all other creatures."

Further defining "search strategies" and "invention" and the elusive characteristics of generativity and creativity that distinguish human cognition might well benefit from the insight of Bateson (1991; 1979; 1972) and others who have begun to outline the representational characteristics and transformational processes of play and how those processes might contribute to what Eccles (1989, p. 240) labels "genetic dynamism... whereby the hominid brain inevitably develops further and further beyond natural selection."

Summary

This essay is, in several senses, a discontinuous one, leaping from head-bone to neck-bone and so on without all considerations due many intervening complications. Let us at least review the bones of the argument to see how they might, in absence of the flesh yet to be added, dance and wiggle in a somewhat appealing way.

Here are the sticks that connect the figure:

Thus, this version of the evolution of representationalism need assume no "stages" in the promulgation of representational skills, but might very well assume some sort of stages in the physical manifestation of those skills in the surrounding environment -- resulting in what appears to be the choppy and discontinuous evolution of the more recent hominid line.

This approach does not attempt to undermine culture-based theories concerning the uses and functions of play. Rather it would attempt to provide the outline for the neurophysiological process of play necessary to assure the continued variety and generativity of human thought. This is, in that sense, an attempt to discover a priori evidence for play -- play as a representational process -- prior to and still beyond all artifacts of culture.

Playing Against the Self: Representation and Evolution
Notes
 

1. This assumes that the mind is most appropriately characterized as an "intentional system" (Dennett, 1981). See Dennett (1981) and Fodor (1990) for a more detailed description of the differences between representational and non-representational philosophies of mind.

2. It has been pointed out to me that the basis of this essay is similar to that of Miller (1973). I agree with this assessment.

Indeed, the conclusions of Miller (1973) -- "we can begin to see that the pleasure of Funktionlust can be as clearly adaptive as the gratification pleasures of eating or sex" (p. 96) -- are strikingly similar to my own. However, the Miller essay does not, as this essay does, deal with the peculiarly dialectic characteristics of representational skills, preferring instead to emphasize the adaptive functions of play as manifest (most often if not always) in the refinement and perfection of existing physical skills. In this respect, Miller's approach is perhaps more similar to that of Fagen (see references cited) than my own.

3. Species variation in Darwinian evolution is a stochastic process, resulting from the random sexual combination and recombination of various genotypes or hereditary potentials. The physical and observable manifestation of these hereditary potentials is called a phenotype; and, to serve as the basis for natural selection, a genotype's phenotype(s) must directly affect the organism's survival potential -- that is, the phenotype(s) in question must either make the organism more or less fit within its natural-historical environment.

This relationship -- between hereditary and survival potentials -- closely parallels the relationship between what I am calling an organism's characteristics and its skills. That is, a phenotype is normally considered in the context of all possible genetically and environmentally influenced expressions of a single genotype. Likewise, each characteristic of an organism entails a large array of potential survival skills. For instance, the long neck of the giraffe allows the giraffe to feed from tall edible-leaf-bearing trees and to cross deep rivers without having to swim -- both potential survival skills. However, the possession of these skills, either singularly or in array, does not entail a singular characteristic. In the case of the giraffe, for instance, these same skills might have resulted from having wings instead of a long neck.

In this essay, for simplicity's sake alone, I will tend to talk about characteristics and skills more often than genotypes and phenotypes. However, this discussion does not attempt to exclude certain behavioral "skills" (instincts) as genotypical, nor certain "characteristics" of organisms as phenotypical, in the broader sense. The idea here is merely to concentrate on the relationship between species characteristics that are genetically determined and species survival skills that result directly from those particular characteristics.

4. This conclusion is similar in most respects to "the law formulated by Copes [that] states that only non-specialized organisms are able to evolve further." (Devillers & Chaline, 1993, p. 163). See also the framework necessary for "organic selection," as defined by James Mark Baldwin and as espoused in Russell (1978) and Sinha (1984).

Connected to this point is another concerning the Hardy-Weinberg principle, which simply states that, in the absence of natural selection, genotype frequencies will remain constant from one generation to the next. Thus, dominant genes (called "normalities" later in this essay) will not displace recessive genes ("abnormalities") during random sexual reproduction.

5. This is, at first glance, in contradiction to the seemingly more popular perspective of species evolution based on the "survival of the fittest" -- rather than, as this example supposes, the increasingly random propagation of the "unfit." But then the unit of analysis in classical evolution theory has always been unclear, ranging in various discussions from gene to community. See the discussion in Hull (1984), pp. 24-30, on different conceptualizations of "species."

6. Even given a social species whose survival potential as a group is enhanced by communication among its individual members (skills that representational characteristics seem particularly suited to provide), there is no obvious need to adopt overly complex representational systems to enhance those communications -- thus the chemicals of the ants and those wiggly dances of the bees.

7. Or, in other words, if we did not play 'against' the self, we would not play at all. And playing then becomes that which is, simultaneously, defined by the self (ie. adapted to self) and outside its boundaries (ie. beyond the limits of the representational skills used to construct self).

8. The argument also, of course, is oriented toward the psychological development of the species and does not directly consider social and culture influences on cognitive evolution. It would be an assumption of this model, however, that social-cultural influences do not qualify as a species-independent mechanism of natural selection and thus serve rather to confine and normalize, rather than extend and abnormalize -- as does the proposed mechanism of play.

Note that also, because I am, in this particular context, limiting the argument to representational skills and, simultaneously, assuming the presence of some extrinsic representational medium (pictorial, linguistic, cultural, etc.) that communicates "successful" representational combinations to subsequent generations, there is somewhat less necessity to justify allusions to a Lamarckian process wherein somatic change induces genetic change.

9. And the particularization of this approach within the boundaries of representionalism distinguishes it from others similarly concerned with the evolutionary function(s) of play. Fagen (1984; 1981), for instance, shares many of the same goals of this essay, but does not define, as this essay does, play most fundamentally as a cognitive, representational, symbol-transformation process and does not, therefore, distinguish cognitive play as an evolution mechanism significantly different from its behavioral manifestations.

Though there are important insights in Fagen's analysis...

...these are interpreted solely in the context of play motivating behavior. Fagen, for instance, hopes for "more tangible manifestation" of "ecological dialectics" that are currently "abstract and philosophical." (1984, p. 167) There is no discussion, as there is here, concerning the representational nature of play and how unavoidably "tangible" manifestations of cognitive play might well tend to bind that play to the non-representational, physical environment of the species and thus necessarily reduce (in order to avoid disastrous physical consequences) the variations of representational skills constructed through play.

10. As already mentioned, Fagen (1984; 1981) is a notable exception in this regard, but his work, though influential and insightful, considers the play of animals in general and does not deal with the apparent discontinuity (see the discussion in Burghardt, 1984, pp. 7-8) between, for instance, reptiles and mammals -- much less among various orders of primates.

Playing Against the Self: Representation and Evolution
References

Bateson, G. (1991). Sacred unity: Further steps to an ecology of mind. New York: Harper Collins.

Bateson, G. (1979). Mind and nature. New York: Dutton.

Bateson, G. (1972). Steps to an ecology of mind. New York: Ballantine.

Bickerton, D. (1990). Language and species. Chicago: University of Chicago Press.

Bruner, J.S. (1972). Nature and uses of immaturity. American Psychologist, 27, 687-708.

Campbell, J.H. (1985). An organization interpretation of evolution. in D.J. Depew and B.H. Weber [eds.] Evolution at a crossroads: The new biology and the new philosophy of science. pp. 133-167. Cambridge: MIT Press.

Corballis, M. (1991). The lop-sided ape. New York: Oxford University Press.

Chick, G. (1989). On the categorization of games. Play & Culture, 2, 283-292.

Csikszentmihalyi, M. & Csikszentmihalyi, I. S. [Eds.] (1988). Optimal experience. New York: Cambridge University Press.

Dennett, D. (1981). Brainstorms. Cambridge, Mass.: MIT Press.

DePryck, K. (1993). Knowledge, evolution, and paradox: the ontology of language. Albany: State University of New York Press.

Devillers, C. & Chaline, J. (1993). Evolution: An evolving theory. New York: Springer-Verlag.

Eccles, J. (1989). Evolution of the brain. New York: Routledge

Eibl-Eiblesfeldt, I. (1967). Concepts of ethology and their significance in the study of human behavior. In H.W. Stevenson, E.H. Hess and H.L. Rheingold [eds.] Early bheavior: Comparative and Developmental approaches. New York: Wiley.

Edelman, G.M. (1989). The remembered present. New York: Basic Books.

Fagen, R. (1984). Play and behavioral flexibility. in Peter K. Smith [ed.] Play in animals and humans. New York: Basil Blackwell. pp. 159-173.

Fagen, R. (1981). Animal play behavior. New York: Oxford University Press.

Fodor. J. (1990). A theory of content and other essays. Cambridge, MA: MIT Press.

Gardner, H. (1983). Frames of mind. New York: Basic Books.

Gazzaniga, M.S. (1992). Nature's mind. New York: Basic Books.

Glansdorff, P. & Prigogine, I. (1971). Thermodynamic theory of structure, stability and fluctuations. New York: Wiley and Sons.

Gould, S.J. (1977). Ontogeny and phylogeny. Cambridge, Mass.: Harvard University Press.

Ho, M.W. & Saunders, P.T. (1979). Beyond neo-Darwinism -- an epigenetic approach to evolution. Journal of Theoretical Biology, 78, 573-591.

Hull, D.L. (1984). Historical entities and historical narratives. in Minds, machines and evolution: philosophical studies. C. Hookway (ed.) pp. 17-42. Cambridge: Cambridge University Press.

Hunter, D.E. & Whitten, P. (1977). Encyclopedia of anthropology. New York: Harper & Row.

Levinton, J.S. (1992, November). The big bang of animal evolution. Scientific Americian. pp. 84-91.

Mayr, E. (1991). One long argument: Charles Darwin and the genesis of modern evolutionary thought. Cambridge: Harvard University Press.

Miller, S. (1973). Ends, means, and galumphing: Some leitmotifs of play. American Anthropologist, 75(1), 87-98.

Myers, D. (1991). Computer Game Semiotics. Play & Culture, 4 (4), 334-345.

Paulson, W. (1991). Literature, complexity, interdisciplinarity. in Hayles, K. N. (ed.) (1991). Chaos and order. pp. 37-53. Chicago: University of Chicago Press.

Piaget, J. (1928). Judgment and reasoning in the child. New York: Harcourt, Brace and Company.

Russell, J. (1978). The acquisition of knowledge. London: Macmillan.

Sinha, C. (1988). Language and representation: a socio-naturalistic approach to human development. New York: New York University Press.

Sinha, C. (1984). Biosocial evolution. in Mae-Wan Ho and Peter T. Saunders [eds.] Beyond neo-Darwinism: An introduction to the new evolutionary paradigm. London: Academic Press. pp. 331--362.

Sternberg, R.J. (1990). Metaphors of the mind: Conceptions of the nature of intelligence. New York: Cambridge University Press.

Sternberg, R.J. (1988). The triarchic mind: A new theory of human intelligence. New York: Viking.

Sutton-Smith, B. (1971). Conclusion. In R.E. Herron and B. Sutton-Smith [eds.] Child's Play. New York: Wiley.

Sutton-Smith, B. & Kelly-Byrne, D.(1984). The idealization of play. in Peter K. Smith [ed.] Play in animals and humans. New York: Basil Blackwell. pp. 305-322.

Trehub, A. (1991). The cognitive brain. Cambridge, MA: MIT Press.

Vygotsky, L.S. (1978). Mind in society: The development of higher psychological process. (M. Cole, V. John-Steiner, S. Scribner, & E. Souberman, eds.) Cambridge: Harvard University Press.

Wicken, J.S. (1984). On the increase in complexity in evolution. in Mae-Wan Ho and Peter T. Saunders [eds.] Beyond neo-Darwinism: An introduction to the new evolutionary paradigm. London: Academic Press. pp. 89-112.