What is the meaning of the concept of "Recognition"?

What is the meaning of the concept of "Recognition"?

Jacques P. BEAUGRAND

Beaugrand.Jacques@UQAM.ca

The present paper is preoccupied by the lack of clarity of the concept of «recognition» which used and abused by ethologists. The definition of a concept has profound epistemological and methodological consequences and critical examination of concepts in public is an essential part in the building of a common scientific language. In so doing, however, one shows attitudes toward other fundamental aspects of the discipline and thus lays oneself open to further criticism. The term "recognition" (e.g. as in «individual recognition», «kin recognition») is used in the literature with (at least) two different connotations. One purports to discriminative overt behaviour or to its consequences (meaning #1). Recognition is also used with a second meaning, to point to some covert brain process which is, explicitly, considered to be behind the observable behaviour. It is important that ethologists clearly establish a distinction between the two uses in their papers and at least attempt to give a definition of what they mean by "recognition" instead of cultivating ambiguity. My suggestion is to reserve the use of the term «recognition» to the brain process. Overt behaviour and its observable consequences (meaning #1) which can be hypothesized to be under the control of the brain recognition (meaning #2) can be qualified in more phenomenal terms, such as «differential treatment» or even simply by the term «behavioural discrimination». Colgan (1983) has treated the various aspects of social «recognition» in animals. Its author equates the concept of «recognition» with the extra-organismal process (meaning #1). «Recognition» is defined operationally by Colgan (1983) as a «concept describing discriminative behaviour related to social categories and not as a theoretical term for some process independent of stimulation and subsequent response» (p. 2). The book is a valuable source of information on behavioural processes which might be taken as evidence for the existence of brain recognitive mechanisms in animals. However, if we except a chapter on sensory processes which probably assists such recognitive brains, the book says nothing on recognition as a brain process.
The term «recognition» is used by many in accord with logical empiricism (or positivism) which is still profoundly influencing ethology and psychology. The definition conforms to descriptive operationalism (a chapter of positivism), which is a semantical and methodological doctrine according to which a concept is synonymous with the corresponding set of operations meant to evidence its material implications (Bridgman, 1927). In the case of recognition, behavioural manifestations of the brain process are (mis-)taken for the brain process itself. In other words, the reference class is assimilated to bodies of evidence or evidence class.
The difference between reference and evidence is best understood in relation to physical or geophysical theories concerning entities which are not accessible to direct observation. For instance, a hypothesis concerning the last glaciation period does not refer to scratches left by glaciers on mountains or to the distribution of plant species today. Scars left by glaciers are not called glaciation by physicists but taken as evidence for it. Similarly, behavioural manifestations should not be confused with re-cognition which is a process occurring in the brain of behaving organisms. They can be considered as indices of recognition and serve as evidence for it.
Not only ethologists should clearly distinguish between reference and evidence of the concepts they use but empirical tests must be preceded by reference analysis. Reference analysis is what we are doing now: to decipher a concept and to define its referent. Without knowing, or at least minimally assuming, what a given statement is about, how can we know how to put it to the test?
Contrary with what some scientists suggest (e.g., Byers & Bekoff, 1986) recognition as a brain process is not necessarily a metaphorical concept. For those who investigate cognitive processes in animals, recognition is a hypothetical concept, just as any mechanism which must be conjectured because it cannot be observed directly. Anyway, ethologists want their explanations and eventual theoretical models to be genuine ones not simply as metaphors. Recognition as a brain process is a scientific hypothesis, to be flourished in the future in the form of models and theories.
Pretending that to collect data in the realm of meaning number one (evidence), and then to make conclusions in the realm of meaning number two (reference) cannot be appropriate is misunderstanding the relation between evidence and hypothesis in science. For sure, evidence must show some degree of relevancy in reference to the concept addressed. Bunge (1967b) marvellously analyses the operation of observing in the science making and, for him, the gathering of evidence can be characterized as purposeful, enlightened and selective perception: purposeful or deliberate because it is made with a given definite aim; enlightened because it is somehow guided by a body of knowledge. It is also selective endeavour because evidence is explored against a background of expectations. Otherwise, the thing looked for would not be recognized when met and would not count as evidence; we would not know how to inquire. It is thus essential that the kind of data that are to count as evidence be stipulated and agreed upon before observation and on the basis of theory or theoretical concepts. Clarification of a concept by working out both its referential and evidential counterparts is thus a requisite before initiating any research. A fortiori, the discussion of methodological aspects for evidencing discrimination and reviewing evidence of recognition is nonsense if it is not preceded by clarification of the concepts. It may happen accidentally that data which was counting as evidence for a given fuzzy concept can be reinterpreted and become evidence for another. In any case, mapping between reference and evidence is better insured in advance than post hoc.
Unfortunately, most experimentalists and naturalists have little faith in conceptual clarification. They rather believe the empiricist myth that facts will speak for themselves and that concepts will finally emerge and be purified by "data". More mature sciences, e.g. physical science, have clearly indicated to us that it is not the case. Conceptualisation, clarification of concepts and theory building must begin as soon as data can be gathered.
One defensive attitude against the effort of conceptual clarification is that to pretend that they have no real referent anyway. For instance, Byers & Bekoff try to disqualify the concept of brain recognition by qualifying it as « black box concept like "drive" ». Their statement is incorrect on two levels. First, recognition as a brain activity is not a black box concept; as mentioned above, it is a scientific hypothesis, to be flourished, to be tested. Second, it does not have the same ontological status as the "drive" concept; for this reason, it cannot be disposed of as easily as "drives" were. These affirmations require further development. Black-boxism is the strategy of modelling systems as empty boxes which respond only to environmental stimuli. Black-boxism can be methodological or metaphysical. Methodological black-boxism is, whether a thing is complex, to conceived it and to treat it experimentally as an empty box with inputs and outputs. Metaphysical black-boxism would add that things are what they appear to be anyway and that any attempt to conjecture their composition and structure is pointless. In conjecturing on the existence of some brainy mechanism in the box one intends to make the box more translucid by tentatively furnishing it with some (software) mechanism called, for the moment, "recognition". So instead of being a "black-box" concept recognition as a brain process might be called a bright- (or at least a grey-) box concept. This attitude has other profound methodological consequences which we will examine later.
But what is the ontological status of "recognition" as a brain process? MacCorquodale and Meehl's (1948) make an interesting distinction between intervening variables and hypothetical constructs, and the distinction is relevant here. An intervening variable is one which relates stimuli to behaviour (responses) without being interpreted in either psychological, psychoneural, cognitive (software) or physiological (hardware) terms; it is an auxiliary symbol solely used to discharge the syntactic or mathematical function. It has no material reality. A hypothetical construct is on the other hand a variable serving as a mediator between some stimulus and behaviour. It is assigned a specific psychological, cognitive or neural interpretation at least hypothetically. In other words, a hypothetical construct is a concept which is supposed to point to something real. The concept of "drive" was problematic because in some contexts it played the role of an intervening variable and in other contexts, the role of hypothetical construct. Moreover, it was a fuzzy concept: whereas some theorists used drive concepts specific to particular types of behaviour, others postulated a "general drive", which was supposed to energize all the patterns of behaviour in an animal's repertoire. It seems that only specific "drives" having the status of hypothetical constructs being supported by what became progressively known from neurophysiological work have survived up to now. Contrary to the old concept of drive, the concept of recognition as a brain process is not fuzzy and has definitively the status of a hypothetical construct, of a scientific hypothesis.

The term «recognition» is used by many in accord with logical empiricism (or positivism) which is still profoundly influencing ethology and psychology. The definition conforms to descriptive operationalism (a chapter of positivism), which is a semantical and methodological doctrine according to which a concept is synonymous with the corresponding set of operations meant to evidence its material implications (Bridgman, 1927). In the case of recognition, behavioural manifestations of the brain process are (mis-)taken for the brain process itself. In other words, the reference class is assimilated to bodies of evidence or evidence class.
The difference between reference and evidence is best understood in relation to physical or geophysical theories concerning entities which are not accessible to direct observation. For instance, a hypothesis concerning the last glaciation period does not refer to scratches left by glaciers on mountains or to the distribution of plant species today. Scars left by glaciers are not called glaciation by physicists but taken as evidence for it. Similarly, behavioural manifestations should not be confused with re-cognition which is a process occurring in the brain of behaving organisms. They can be considered as indices of recognition and serve as evidence for it.
Not only ethologists should clearly distinguish between reference and evidence of the concepts they use but empirical tests must be preceded by reference analysis. Reference analysis is what we are doing now: to decipher a concept and to define its referent. Without knowing, or at least minimally assuming, what a given statement is about, how can we know how to put it to the test?
Contrary with what some scientists suggest (e.g., Byers & Bekoff, 1986) recognition as a brain process is not necessarily a metaphorical concept. For those who investigate cognitive processes in animals, recognition is a hypothetical concept, just as any mechanism which must be conjectured because it cannot be observed directly. Anyway, ethologists want their explanations and eventual theoretical models to be genuine ones not simply as metaphors. Recognition as a brain process is a scientific hypothesis, to be flourished in the future in the form of models and theories.
Pretending that to collect data in the realm of meaning number one (evidence), and then to make conclusions in the realm of meaning number two (reference) cannot be appropriate is misunderstanding the relation between evidence and hypothesis in science. For sure, evidence must show some degree of relevancy in reference to the concept addressed. Bunge (1967b) marvellously analyses the operation of observing in the science making and, for him, the gathering of evidence can be characterized as purposeful, enlightened and selective perception: purposeful or deliberate because it is made with a given definite aim; enlightened because it is somehow guided by a body of knowledge. It is also selective endeavour because evidence is explored against a background of expectations. Otherwise, the thing looked for would not be recognized when met and would not count as evidence; we would not know how to inquire. It is thus essential that the kind of data that are to count as evidence be stipulated and agreed upon before observation and on the basis of theory or theoretical concepts. Clarification of a concept by working out both its referential and evidential counterparts is thus a requisite before initiating any research. A fortiori, the discussion of methodological aspects for evidencing discrimination and reviewing evidence of recognition is nonsense if it is not preceded by clarification of the concepts. It may happen accidentally that data which was counting as evidence for a given fuzzy concept can be reinterpreted and become evidence for another. In any case, mapping between reference and evidence is better insured in advance than post hoc.
Unfortunately, most experimentalists and naturalists have little faith in conceptual clarification. They rather believe the empiricist myth that facts will speak for themselves and that concepts will finally emerge and be purified by "data". More mature sciences, e.g. physical science, have clearly indicated to us that it is not the case. Conceptualisation, clarification of concepts and theory building must begin as soon as data can be gathered.
One defensive attitude against the effort of conceptual clarification is that to pretend that they have no real referent anyway. For instance, Byers & Bekoff try to disqualify the concept of brain recognition by qualifying it as « black box concept like "drive" ». Their statement is incorrect on two levels. First, recognition as a brain activity is not a black box concept; as mentioned above, it is a scientific hypothesis, to be flourished, to be tested. Second, it does not have the same ontological status as the "drive" concept; for this reason, it cannot be disposed of as easily as "drives" were. These affirmations require further development. Black-boxism is the strategy of modelling systems as empty boxes which respond only to environmental stimuli. Black-boxism can be methodological or metaphysical. Methodological black-boxism is, whether a thing is complex, to conceived it and to treat it experimentally as an empty box with inputs and outputs. Metaphysical black-boxism would add that things are what they appear to be anyway and that any attempt to conjecture their composition and structure is pointless. In conjecturing on the existence of some brainy mechanism in the box one intends to make the box more translucid by tentatively furnishing it with some (software) mechanism called, for the moment, "recognition". So instead of being a "black-box" concept recognition as a brain process might be called a bright- (or at least a grey-) box concept. This attitude has other profound methodological consequences which we will examine later.
But what is the ontological status of "recognition" as a brain process? MacCorquodale and Meehl's (1948) make an interesting distinction between intervening variables and hypothetical constructs, and the distinction is relevant here. An intervening variable is one which relates stimuli to behaviour (responses) without being interpreted in either psychological, psychoneural, cognitive (software) or physiological (hardware) terms; it is an auxiliary symbol solely used to discharge the syntactic or mathematical function. It has no material reality. A hypothetical construct is on the other hand a variable serving as a mediator between some stimulus and behaviour. It is assigned a specific psychological, cognitive or neural interpretation at least hypothetically. In other words, a hypothetical construct is a concept which is supposed to point to something real. The concept of "drive" was problematic because in some contexts it played the role of an intervening variable and in other contexts, the role of hypothetical construct. Moreover, it was a fuzzy concept: whereas some theorists used drive concepts specific to particular types of behaviour, others postulated a "general drive", which was supposed to energize all the patterns of behaviour in an animal's repertoire. It seems that only specific "drives" having the status of hypothetical constructs being supported by what became progressively known from neurophysiological work have survived up to now. Contrary to the old concept of drive, the concept of recognition as a brain process is not fuzzy and has definitively the status of a hypothetical construct, of a scientific hypothesis.

A theoretical framework for individual recognition

Recognition can be considered as a special cognitive process. Bunge (1979, 1980, 1985) can be used as  a basis for the elaboration of such a theory of recognition, to which the reader is referred as primary source. The framework of such a theory can only be briefly outlined here.

1. Recognition of an object, be it a given conspecific individual, is identical to events occurring in the brain of animals, not in their behaviour. Behaviour (which is by definition overt) can be taken as evidence for recognition but not assimilated to it.
2. The brain of animals capable of recognition is also capable of detection, sensation (sensory process or feeling), perception and experience and of action.
3. The behaviour of every animal endowed with a nervous system is produced (controlled, mediated, subserved) by the latter.
4. Some of the inherited capabilities of an animal endowed with plastic neural systems are modifiable by learning (e.g. bird song). The part of the behavioural repertoire of an animal which is controlled by the genetically preprogrammed part of its nervous system is called its inherited repertoire (instinctive, stereotyped, modal or rigid), and the one controlled by the plastic part (derived from the genetically preprogrammed part) of its nervous system, its learned repertoire.
5. Include here Bunge's definition of drive given above.
6. Animals have a system of bio-values built into them, whether from birth or shaped during life through learning. A value system for an animal exists if the animal can detect any a given item (thing, state, or event) and discriminate it from all other items, and if for any two items, one of them is preferred. All animals have the ability to prefer some things over others and to avoid harmful stimuli. Even bacteria can swim towards nutrients and away from toxic substances. We can postulate that all animals are capable of evaluating certain internal and environmental states, events. In general, the preference of an animal for a stimulus over another can be explained by the greater adaptive value of the former over the latter. However, preferences are manifestations of brains which were shaped by evolution, not manifestations of evolution. Again, a difference has to be noted between preference and choice: the former motivates the latter in the sense that choice is overt evaluation, preference or valuation in action.
7. Learning is a modification of neural activity. From Bunge (1980) we borrow the following definition of learning: Call E a kind of event or process in a neural system involving a plastic subsystem of an animal a, and S a kind of stimuli (external or internal) which a can sense or detect. Then a has learned e  E in the presence of s  S during the time interval [t1, t2] if and only if (i) e did not occur in a in the presence of s before t1; (ii) after t2, e occurs in a whenever a senses s (i.e. a has memorized s).
8. From this, an animal can expect or foresee a future event of kind E when sensing an external or internal event while in state t, if the animal has learned to associate s and t with an event of kind E. The action of an animal can be said to be purposive because the animal b has learned that doing X brings about, or increases the chances of attaining Y; that b expects the possible occurrence of Y upon doing X; and because b values Y.
9. Moreover, some animals are capable of at least minimal concept formation, i.e. the forming of simple empirical categories. For instance, pigeons can recognize objects presented under unfamiliar perspectives (Hollard & Delius, 1982), and to distinguish slides with people in it from those without (Herrnstein & Loveland, 1964; Malott & Sidall, 1972). Images of two conspecifics in various positions were shown to cocks by Ryan (1982), and they could distinguish new images of the same individuals from equivalent images of another unfamiliar individual. Also, budgerigars can be trained to recognize familiar conspecifics in colour slides they have never seen before (Trillmich, 1976). These findings suggest rather strongly that birds can form concepts of stimulus objects and, a forteriori, of conspecifics. Similar experiments with monkeys attest that primates can conceive rather elaborated categories (Weiskrantz, 1985). Still, a bacterium, by moving away from noxious stimuli of various types lumps them into a single category "bad" and this suggests that forming categories is a very basal process in the animal kingdom. After all, computers are also able to do it.
10. Within this general framework, individual recognition can be defined as a neural event occurring in or produced by the brain of an individual organism (of a certain species having attained a minimum level of nervous plasticity) consisting in (i) discriminating a (usually) conspecific individual, on the basis of a configuration of simple cues, from a background composed of at least one other conspecific, and to (ii) react neurally in a very specific way by instantiating or activating information which was previously (learned to be) associated to that peculiar configuration, and to categorize it into a category of which it is the only member.
11. If the category can include many members, then either there is confusion (mistakes of identification) or generalization, or it is categorization of a larger class of objects (i.e. conspecifics, kin, potential mates, potential predators, humans, &c).
As conceived here, individual recognition is not different from any other recognitive process involving simple cues (Barnard & Burk, 1979). However, the configuration of these simple cues is so peculiar that it points to a pigeon-hole which happens to lodge a unique member. If it lodged a whole class of conspecifics having genes in common (or having a common odour, taste, pheromone, colour/markings) then it might be called kin recognition.
Is it mandatory here to let emergent properties pop out of the philosophical tool-box ? An emergent property is a property which is more than (the sum of) its simple cues, just like a cube shows more properties when compared to the mere sum of its six square surfaces. Arrangement or structure (spatio-temporal relations) between the parts (simple cues) contribute to the emergence of further global and new properties. Moreover, in being the only member of a conceptual category the individual recognized individually receives an additional (emergent) attribute. But there is nothing mysterious about that.
When an individual animal meets another one which it recognizes (and most probably is recognized by), the neural representation of the recognized conspecific is also an instantiation of past states which have been (learned to be) associated to that peculiar configuration, as well as the adequate path of action. In other words, the recognizing individual also remembers past experiences or previous relations which were bound to the recognized individual, including appropriate actions. For instance, on being approached by a previously dominant individual which it recognizes, a subordinate also remembers, for instance, the degree of "danger" bound to it, and/or the adequate action to be taken to avoid it, or to reduce danger. A long tradition of research on memory and learning mechanisms indicates that a memory which is not refreshed on a regular basis, or left in disuse, gradually decays with time. Also, it is well established that the original associative strength is a complex function of the past probability of co-occurrence of the components part of the perceived configuration, of its attributed biovalue (reinforcement), and of the degree of attention which presided to the association.
This primary evocation once available can be used by the cognitive brain to elaborate action, i.e. behaviour program. However, the specific action which is to be taken will depend on several additional conditions related to present motivational state, and to context. Such a restriction is made to account for several regular patterns which could not be accounted for if the reaction path associated with the recognized configuration was systematically put to execution. For instance, Xiphophorus when deprived of food for 24 h, re-acted toward a menacing rival in a quantitatively different manner than when not deprived. Similarly, a familiar dominant or subordinate approaching while holding a threat posture is treated differently than an apparently pacific one. As mentioned above, re-action will also depend on the recency of the last contact. Challenges directed by a previous subordinate toward its previous dominant fish increase in frequency as a direct function of the period of separation. In addition, after 48 h of separation, an intermediate experience of dominance given to the subordinate increases challenges addressed by the latter to its previous dominant when reintroduced into their home aquarium. However, the polarity of the reinstated dominance-subordination relation is not significantly changed from the initial one (results from yet unpublished experiments).
As a general rule one can say that if motivational and contextual conditions have not changed since the last strengthening of the cross-modal association, then the path of action which was originally the most strongly associated to the configuration is the one most likely to be followed. But this varies according to the species studied.
In Xiphophorus, a non territorial species, the spatial context in which conflict occurs does not seem to be relevant for familiar pairs. This is the case when two fish which had the opportunity to establish a dominance relation in the first place meet again after a separation of 3, 6, or 24 hours. As first evidenced by Zayan (1974, 1975), confirmed by Beaugrand & Zayan (1985), the initially established relationship was immediately reinstated, even if the previously alpha individual met its previous omega partner in the latter's residence tank. The fish rapidly confirmed their initial  dominance relationship. They "knew" each other and familiarity with the opponent was definitively more determinant on conflict issue than familiarity with the milieu. When compared to unfamiliar control pairs having received equivalent  or alpha divergent social experiences, familiar pairs manifested almost no reciprocal aggression.
In "territorial" species, the spatial context in which the conflict occurs is critical. The polarity of the dominance relation which is originally established between two fish is not strictly determinant of all future encounters between the now familiar individuals. It depends on where the pair encounters. For instance, De Boer & Heuts (1973) have shown that territorial defence could be induced at will in the fish Hemichromis bimaculatus by alternatively provoking intrusion into each of two familiar fish's residence tanks. However, in yet unpublished experiments on the same fish species, the present author found that if an initial dominance relation was established in an unfamiliar tank, it maintained the same polarity when the now familiar fish pair was transferred into successive unfamiliar tanks. Again, when compared to unfamiliar pairs repeatedly staged to encounter in unfamiliar tanks, drastic differences in agonistic behaviour could be found between familiar and unfamiliar pairs, strongly supporting the hypothesis that individual recognition was nevertheless occurring in the residence situation. The fact that individual recognition was happening while not materially implying stability of the original dominance relation is disturbing for people relying on dominance reversal or stability as their only valuable indicator of recognition. Again, these results stress the importance of distinguishing both conceptually and methodologically individual recognition as a hypothetical process and keeping it distinct from the behavioural process which can serve as evidence for the former.
The brainy process of individual recognition just superficially conjectured remains, after all, a pure ghost. It will always remain hidden from the observations made by the ethologist, just like any hypothesis related, for instance, to natural selection or to the origin of epigamic displays. After all, it is the role of theories to account for the observable, using for this task unobservable entities or processes conjectured as situated «behind» what is to be explained. The present framework concerns cognitive competence. To be tested, it has to be tied up to the "evidence" or "performance" side of the ethological reality, a task which has already begun in the last paragraphs. Behavioural indicators of individual recognition have been amply scrutinized by Colgan (1983) and by Zayan (1987a, 1987b), to which authorities the reader is referred.
The main requisites for the formulation (not yet for their acceptance) of scientific hypotheses seem to be satisfied by the framework of brainy recognition: (i) the hypothesis is formally correct (it does not contain self contradictions); (ii) the hypothesis is semantically non empty in some scientific context; (iii) the hypothesis is grounded to some extent on previous knowledge and compatible with the bulk of scientific knowledge in the domain of neuropsychology, learning and memory, and cognition; (iv) the hypothesis is empirically testable by the objective procedure of science, i.e. by confrontation of its derived consequences with empirical data controlled in turn by scientific techniques and theories.
Before closing the present comment, I would like to stress another point which concerns the final mise-en-garde of Byers & Bekoff (1986) as regards the "fallacies associated with arbitrary inference concerning processes about which we know little or nothing" (p. 344). Theoretical poverty which characterizes the behavioural sciences has always been, in this way, extolled as a virtue, when it came to obey the credo of positivism. Inference is absolutely necessary in order to gain more profound knowledge on what is controlling or causing what can be readily observed. Remember, behaviour is the thing to be explained by ethologists. Behaviour cannot account for, or explain behaviour. Deeper support is required for this task.
« Scientific knowledge is data cooked with the help of theory » (Bunge, 1967b). The recent and impressive growth of knowledge about the evolutionary aspects of behaviour attest to this. This was made possible by the spreading in ethology of theories and models developed by border sciences such as ecology, population biology and sociobiology. These concepts and theories were used to account/explain ethological regularities and to look forward to new ones. Conceptualisation and theorizing have in this manner played an important role in regard to the study of the origin of behaviour, and nowadays ethologists do not hesitate to speculate on evolutionary "causes" of behaviour, without seeing the spectre of strict operationalism agitated each time.
The analysis of evolutionary processes that could have assisted in the selection of behavioural capacities in animals is, by nature, a very speculative topic. Nevertheless, it has become fashionable in ethological literature to present evolutionary hypotheses with great assurance. Why should the intellectual activity consisting in hypothesizing on animal cognition be considered as mere speculation and loss of time, whilst thinking aloud on evolutionary mechanisms as essential for bringing the daily bread ? Do we know much less on basic brain and cognitive mechanisms than on evolutionary factors that could justify this differential, not to say «schizoid» attitude: total reserve (both substantive and methodological) on the former topic and generosity of ideas (plus tolerance) on the latter topic ?
Let us, or let others also have the audacity to speculate loud and clear on brain and cognitive mechanisms (possibly) controlling animal behaviour. Ethology will surely benefit from a further cross-fertilization with the neuro- and cognitive sciences. After all, proximate causation of behaviour, which is also part of the original program of ethology, was quite a neglected chapter in recent years.
Finally, let me conclude by a quotation which I feel is quite à propos. «The use of a vocabulary within a particular field at a given moment goes far beyond a mere question of words; it is an indicator of both the will to circumscribe with a maximum of accuracy the specific problems treated, and the major trends of interests of the scientists concerned » (Thinès, 1987, p. 2).

REFERENCES

BARNARD, C.J. & T. BURK, 1979: Dominance Hierarchies and the Evolution of "Individual Recognition". J. Theor. Biol., 81, 65-73. 
BEAUGRAND, J. & R. ZAYAN, 1985: An experimental model of aggressive dominance in Xiphophorus helleri (Pisces, Poeciliidae). Behav. Proc., 10, 1-52. 
BRIDGMAN, P.C., 1927: The logic of modern physics. Macmillan, New York. 
BUNGE, M., 1967a: Scientific research I. The search for system. Springer‑Verlag, New York.
BUNGE, M., 1967b: Scientific research II. The search for truth. Spinger‑Verlag, New York.
BUNGE, M., 1979: Treatise on Basic Philosophy, vol. 4: Ontology II. D. Reidel, Boston.
BUNGE, M., 1980: The Mind-Body Problem. A psychobiological Approach. Pergamon, New York.
BUNGE, M., 1985: Treatise on Basic Philosophy, vol. 7: Epistemology and Methodology III. D. Reidel, Boston.
BYERS, J.A & M. BEKOFF, 1986: What does "Kin Recognition" Mean ? Ethology, 72, 342-345.
COLGAN, P., 1983: Comparative Social Recognition. Wiley, New York.
DeBOER, J.N. & B.A. HEUTS, 1973: Prior exposure to visual cues affecting dominance in the Jewell fish, Hemichromis bimaculatus (Pisces, Cichlidae). Behaviour, 44, 299-321.
HERRNSTEIN,R.J. & D.H. LOVELAND, 1964: Complex visual concepts in the pigeon. Science, 146, 549-551.
HINDE, R.A., 1970: Animal Behaviour. A synthesis of Ethology and Comparative Psychology, 2nd ed. McGraw Hill, New York.
HOLLARD, V.D. & J.D. DELIUS, 1982: Rotational invariance in visual pattern recognition by pigeons and humans. Science, 218, 804-806.
MacCORQUODALE, K. & P.E. MEEHL, 1948: On a distinction between hypothetical constructs and intervening variables. Psychol. Review, 55, 95-107.
MALOTT, R.W. & J.W. SIDALL, 1972: Acquisition of the people concept in pigeons. Psychol. Report, 31, 3-13.
MCBRIDE, G., 1971: Theories of animal spacing: the role of flight, fight and social distance. In: Behaviour and Environment (A. H. ESSER, ed.). Plenum Press, New York, pp 53‑68.
RYAN, C.M.E., 1982: Concept formation and individual recognition in the domestic chicken (Gallus gallus). Behav. Anal. Letters, 2, 213-220.
THINES, G., 1987: The scientific assessment of animal subjective experiences: from Buytendijk's animal psychology to animal welfare. In: Cognitive Aspects of Social Behaviour in the Domestic Fowl. (ZAYAN, R. & I.J.H. DUNCAN, eds.). Elsevier, New York, pp. 2-18. 
TRILLMICH, F., 1976: Learning Experiments on Individual Recognition in Budgerigars (Melopsittacus undulatus). Z. Tierpsychol., 41, 372-395. 
WEISKRANTZ, L., 1985: Introduction: Categorization, cleverness and consciousness. Philo. Trans. Roy. Soc. London, B298, 3-19. 
ZAYAN, R., 1974: Le rôle de la reconnaissance individuelle dans la stabilité des relations hiérarchiques chez Xiphophorus (Pisces, Poeciliidae). Behaviour, 49, 268-312. 
ZAYAN, R., 1975: Défense du territoire et reconnaissance individuelle chez Xiphophorus (Pisces, Poeciliidae). Behaviour, 52, 266-312. 
ZAYAN, R., 1986: The epistemological functions of an Ethological Model. In: Relevance of models and theories in ethology. (CAMPAN, R. & R. ZAYAN, eds). Privat, I.E.C. Toulouse 1986, Vol. 1, pp. 25-45. 
ZAYAN, R., 1987a: Recognition between individuals indicated by aggression and dominance in pairs of domestic fowl. In: Cognitive Aspects of Social Behaviour in the Domestic Fowl. (ZAYAN, R. & I.J.H. DUNCAN, eds.). Elsevier, New York, pp. 321-438. 
ZAYAN, R., 1987b: Recognition of individuals indicated by activity independent of agonistic context in laying hens. In: Cognitive Aspects of Social Behaviour in the Domestic Fowl. (ZAYAN, R. & I.J.H. DUNCAN, eds.). Elsevier, New York, pp. 439-492.