Indexic properties

An index carries information by literally pointing out its referent, so one may ask how optical signals might be constructed to act as spatial pointers. In terms of signal-elements (table 4-III) the question becomes whether orientations, movements, shapes, structures and reflectance patterns may be used to index referents.

The mere body orientation of a sender is widely used by animals as an indexic signal pointing out various referents. The commonest referent is the social companion this is being addressed (the intended receiver). The orientation adopted by the sender to point out the addressee may vary according to the anatomy of the communicants, to the other information being transferred at the same time and to many other factors (e.g., table 7-V). In fact, pointing out the addressee is also a syntactic problem, to be considered below.

Orientation with respect to referents other than the addressee is also common. For example, an animal may give some particular signal when sighting a potential predator, and then visually track the predator sighted. Social companions may then look at the sender's direction of gaze in order to locate the predator, so that the direction is indexic (e.g., Hall and DeVore, 1965; Struhsaker, 1967). One may expect accompanying morphological structures or color patterns to be arranged so as to make this indexic signal more conspicuous and unambiguous. It seems possible that the head-striping of many small songbirds serves such accentuation of the indexic direction of gaze, since regardless of a viewer's position relative to the bird, the directions its head is turned becomes quite obvious (fig 8-1a).

Fig 8-1. Coloration to enhance indexic orientations. The convergent longitudinal markings on the head of the lark sparrow (a) facilitate recognition of its direction of gaze, and the markings on the pinnae of jackrabbits (b) facilitate detecting their orientation with respect to a sound-source.

It is largely irrelevant whether the sender fixes the object of gaze monocularly or binocularly, so long as the receiver knows which and can reconstruct the probable direction of gaze from seeing the position of the sender's head. If stripes were to run from one side of the head over the cap to the other, instead of running longitudinally to converge at the bill, the indexic information would be more difficult to recover; such a pattern, which would be disruptive (ch 6), is rare among birds. Other potential functions of head-striping in birds were discussed in ch 4 in relation to hiding the eye, preventing glare, providing sighting lines,etc.

Years ago, Haartman (1957) drew attention to indexic signals of hole-nesting birds, which display at the nestsite. He noted that birds such as woodpeckers and nuthatches tend to have dorsal signal-markings, since the backs of these birds are visible to mates when they are pointing out possible nesting holes.

S. Witkin (pers. comm.) notes that in hole-nesting and in other birds presumed indexic coloration is frequently triangular in shape. For example, hole-nesting chickadees (Parus) have a prominent white triangle on the cheek, with the apex pointing toward the bill. The head, as seen from above, displays a similar black triangle, and the chin as seen from below the bird also displays a black triangle. The dark head-patterns of many other birds when seen from above also form a triangle pointing toward the bill. Therefore, triangularity (as well as the kind of striping illustrated in fig 8-la) may be an important indexic color pattern in animals.

Orientation of an appendage, as opposed to gross orientation of the head or body, sometimes constitutes an indexic signal. The commonest example is the orientation of the pinnae in various mammals: the pinnae are oriented toward sounds, particularly unusual sounds that might connote danger, and so a companion viewing the listening animal may be able to approximate the direction of the sound even if he himself did not hear it. It is common to find special coloration on mammalian pinnae (e.g., black tips on the ears of lagomorphs, fig 8-lb) , although like headstriping in birds these marks may be involved in other communication as well. In some cases, special structures, such as the eartufts on some felids, also accentuate the indexical quality of pinnae-orientation (see figure 8 in Hailman, 1977a).

Movement, by definition, involves a change in position so that the directional components of movements could serve as indexic signals. For example, the response of some social animals is to flee a potential danger, so that the direction of fleeing indexes the direction of danger as being immediately behind. In such cases one expects posterior, rather than anterior, enhancement by color pattern, and this is commonly the case (e.g., the white-tailed deer's white tail is raised when fleeing: see figure 20 in Hailman, 1977a).

Movement of part of an animal, as opposed to total body displacement in space, may also be indexic. Statements by Lorenz (e.g., 1941) to the effect that the inciting movement of female mallards is rigidly fixed and unoriented notwithstanding, quantitative study shows clearly that inciting is an indexic signal (Stillwell and Hailman, in prep.). The female turns her head toward an intruding male, stopping the movement when looking binocularly directly at him, then returning her head to straightforward and repeating the directed inciting motion (fig 8-2a). It is instructive to note that, except for usually hidden wing-markings, the female mallard is nearly uniformly a concealing brown color. The one patch of signal-coloration visible while swimming is the orange-and-black marking of her bill--the pointer used in the inciting signal.

Fig 8-2. Indexic signals include movements, such as the female mallard's inciting (a), in which her orange bill is moved to point toward an intruding male; shapes, as in pointing with the index finger (b); structure, as in the gill covers of certain male fishes (c) that index the position of the female for spawning; and color-patterns, such as the markings on the anal fins of male mouth-brooding cichlids (d) that direct the female to the proper place to take the male's sperm into her mouth.

The shape of an animal is a behavioral signal-element created by adjustment of body parts relative to one another (ch 4). Although orientation of the head (e.g. fig 8-1a) is strictly speaking a kind of behaviorally created shape, ethologists usually separate such cases from more stereotyped postures that create less common shapes. For example, the open hand is not immediately taken by most persons as an optical signal, unless its orientation toward someone indicates it is indexic greeting. However, a fist with the index finger protruding is a special shape that is immediately taken to be an indexic signal (fig 8-2b); the receiver confirms or refutes this communicative hypothesis by looking in the direction toward which the finger is pointing to see if something special is there. Indeed, pointing with the index finger is so familiar that it might be called our archetype optical index, yet such signals seem surprisingly rare in animals. Chimpanzees also use a finger indexically, although perhaps only after contact with humans. Hunting dogs appropriately called pointers index prey by a special body posture in addition to the orientational component of their direction of gaze.

In sum, behavioral elements used as indexic signals are often accompanied by patterns of coloration that enhance the indexic quality. Examples are summarized in table 8-1.

Special structures evolved for indexic functions in animal communication seem rare, although as noted above structural adaptations may enhance the indexic qualities of orientations, movements and postures. Nelson (1964) notes that the glandulocaudine fish Corynopoma has evolved a special “courtship paddle" that is extended for the female to nip at (fig 8-2c). The position of the end of the paddle in a sense indexes the proper spatial position of the female for spawning relative to the male, and hence is a special indexic structure.

Finally, there are many examples of indexic colorpatterns of animals, apart from coloration to enhance orientation, shape and movement elements of behavior. For example, in some mouth-brooding cichlid fishes the young fry return to the dark gape of the parent and when older are attracted to stay near the closed-mouthed parent by attempting to enter a dark spot on its side. The spot is thus initially a deceptive mechanism inducing the young to stay close while they presumably learn optical, chemical and other characteristics of the parent. Another example from cichlid fishes is the marking on the male's anal fin in Haplochromis burtoni (Wickler, 1968). After laying eggs, the female takes them into her mouth, and the male lies near the substrate while emitting sperm. The female puts her mouth near the indexic fin-spots and ingests the sperm for fertilization of the eggs in her mouth cavity (fig 8-2d , p. 253). Wickler believes the spots to be egg-mimics that deceive the female, who attempts to pick up the "eggs" and thereby picks up sperm; however, this interpretation is not the only one possible (see ch 9).

Table 8-I

Behavioral Indexes

In sum, indexic signals are widespread in animals, being found in both behavioral and morphological elements of optical signals of all kinds. Although not mentioned specifically here, extrinsic signals also may be indexic, and examples are provided later in this chapter. In some cases, knowledge of how indexic signals work leads to expectations concerning color patterns that enhance the indexic qualities (table 8-I) , and it would seem fruitful to give such patterns more analytical attention in future studies.

Iconic properties

An iconic signal carries information by physically resembling its referent, so one may ask how optical signals might be constructed to resemble other things. Although they were not identified by name as iconic, many of the deceptive signals cited in ch 6 are iconic in nature: all mimicry, for example, is iconic at least in a derivative sense. For example, if a bird pecks at a moth, which then spreads its underwings to reveal two large, dark spots, then the moth is said to be mimicking the eyes of some large animal that would frighten the bird. The spots are an icon for eyes because they look to us like eyes, although the bird presumably mistakes them for eyes. Therefore, it becomes useful to separate deceptive mimicry (signals actually mistaken for other things) from general mimicry or iconic signals in the restricted sense (signals recognized as signals, which look like something else).

Iconic behavioral signals (especially movements and postures) are almost universal in animal optical communication, although one might not readily recognize this fact. When a gull, for example, adopts the upright posture (the "threat" illustrated in figure 24 of Hailman, 1977a) the signal looks like the start of a peck at the opponent and has probably evolved from just such a behavioral pattern. When the gull pecks into the grass, the movement looks like fighting, even though it does not involve an opponent; ethologists call this "redirected aggression." Such redirected movement is an icon to the opponent on the adjoining territory, another kind of "threat" signal. In these cases of iconic signals, the referent is the behavior that the sender might engage in subsequently, say if the opponent were to move closer to the sender. Marler (e.g., 1977) has called such icons "predictors." It is generally believed among ethologists that the overwhelming majority of behavioral signals are evolved from intention movements of non-communicative behavior (see references in Hailman, 1977a and table 4-IV, p. 107). Therefore, all such signales are probably iconic to some extent--they look like other behavior that the sender is broadcasting as probable in the immediate future.

It is useful to point out that iconic behavioral signals do not all have to evolve through the process of phylogenetic ritualization. The form of iconic signals may also develop through experience of the individual sender (table l-II). A boy may learn, for example, that clenching his fist is noticed by a playmate as preparation for hitting; after that, the boy may purposely close his fist in order to communicate, consciously or subconsciously, his willingness to fight. In both phylogenetic and onto genetic developments of iconic signals, the receiver already cues upon preparatory or "intention" movements of some act; then natural selection or individual experience can proceed to enhance the signal-value of the icon to make it more conspicuous and unambiguous.

The question is therefore not whether individual movements and postures can be icons, since many of them clearly are. Rather, the question is whether orientations, morphological specializations and other signals such as whole complexes of behavioral signals are also iconic. In the case of orientation, it is difficult to find an example of a pure icon. Orientation, say toward the addressee, may accompany an iconic movement or posture, but in this case it is only indexic--unless one cares to argue that the orientation of the icon is itself iconic for the orientation of the icon's referent. For example, a threat posture may be an iconic signal for possible attack and the orientation of the posture toward the opponent indexes the intended receiver of the signal. Since the receiver is also the intended recipient of attack, the orientation of the signal itself may be taken as an icon for the directedness of the attack.

Structures and color-patterns are frequently iconic. Many mammals have special warts or other structures that look like fighting weapons such as teeth or horns (see figure 7 in Hailman, 1977a), and others have color-patterns on the ears or elsewhere that also look like weapons (see figure 22 in Hailman, 1977a). These structures and patterns have been called "automimetic," although it is doubtful that they are truly deceptive in the sense of ch 6. Opponents probably do not perceive these iconic signals as actual additional weapons. However, because opponents do recognize the display of weaponry itself as signaling probable fighting, the repetitious nature (table 7-I) of the added weapon-icons enhances the signal by increasing redundancy.

There is an interesting reputed case of "automimicry" in which coloration creates the iconic resemblance, but the color-pattern itself is unexplained. The male mandrill has a red nose and blue cheek skin that resembles the red penis and blue scrotal area. Wickler (1968) cites primate examples in which anogenital display is used in social, rank-related signaling, including cases in which males develop female-like swellings that are used as social signals. Presumably the mandrill's face mimics genital signals, in effect repeating them anteriorly for use in face-to-face signaling, but the coloration of neither genital nor facial areas has been well explained.

Jouventin (1975) calls the coloration “luminous,” and points out that it is conspicuous in the thick equatorial forest. This possibility is supported by the fact that red and blue are both complementary to green colorations (fig 7-3,p. 203), and hence might contrast well against various backgrounds of green vegetation. Another possibility is that the coloration makes use of the depth-illusion based on color (fig 5-18, p. 152), and hence enhances the basal relief of the signal-morphology. Because red is seen as closer than blue or violet, the penis may appear to stand out from the blue scrotal background. A blue object would not protrude from a red background, so the hypothesis provides a clear prediction concerning coloration to enhance relief of morphological signals.

Extrinsic signals are natural candidates for iconic representations, but they seem rare in most animals. Every drawing of an animal in this book is an icon, but animals themselves rarely draw pictures. Many iconic-like signals of animals appear to involve true deception. For example, male of some species of empiid flies present prey to the female before copulation, which occupies her sufficiently that she may not eat the male. In other species, the male first wraps the prey in silk, and in still others presents an empty silk balloon (Kessel, 1955). In the last case, the female may be deceived and actually search for prey, but it seems more likely that the balloon has become symbolic of the male's copulatory intentions; by either interpretation, though, the balloon is not really an iconic signal. In deceptive mimicry there are examples such as the Asiatic spider that builds spider-like masses in its webs (fig 6-4). Like moth eye-spots, however, the receiver actually mistakes the stimulus for something else rather than perceiving it as a signal that stands for something else.

Waxing speculative, a fascinating bit of natural history might hold an example of an extrinsic icon. The garibaldi is a Pacific fish growing to about 30 cm in which the male is bright orange. He holds a year-round territory on to which he attracts females during the breeding season, partially by culturing a 15- to 40-cm oblong patch of red algae of various species. The patch is thus about the same size as the fish, and the patch acts as a nestsite, with eggs being attached to the algae by short filaments (T.A. Clarke, 1970). Is the male's coloration evolved as an iconic signal relating to the nest-site, or is the nest-site an icon of the male? Or is the similarity of shape, size and color purely evolutionary coincidence?

symbolic and mixed properties

A symbolic signal carries information by "standing for" its referent in any manner held in common by sender and receiver. For present purposes it may be taken as any relation between signal and referent that is neither indexic nor iconic. Sometimes symbols are called arbitrary, but I have avoided that word as being ambiguous.

"Arbitrary" means established by caprice rather than reason, and in most cases one does not know how symbols became established in animals. The actual relation between a symbol and its referent may be arbitrary in the sense of having no specifiable connection (such as the signal pointing out or resembling the referent), but there may be good historical reasons for having established the relation originally. For example, the symbol "π" for the ratio of the circumference of a circle to its diameter is "arbitrary" in the sense that it bears no obvious relation to the words quoted as its referent, to a drawn circle, or even to the numerical value of the ratio itself (3.14159+). However, the relationship was not established arbitrarily. The symbol is the 16th letter of the Greek alphabet, the initial letter of peripherion (periphery), and hence is very nearly a simple abbreviation. Analogously, many of the various symbolic elements of behavioral signals of animals probably evolved from behavior precursors that are obscure today. The signals are neither indexes nor icons, hence they are symbols, but their historical establishment was not necessarily arbitrary in any useful sense of the word.

It is evident that many behavioral elements of signals (orientations, postures and movements) are symbolic, as are many morphological elements (structures and colorpatterns), since the only operational criterion is that they be neither indexic nor iconic. For example, the adult male common flicker has a mustache mark lacking in the female (figure 10 in Hailman, 1977a), so that the presence or absence of this signal symbolizes the sex of the bird. Whenever the referent is something that cannot be pointed our in a literal sense or represented iconically, the signal that stands for it must be a symbol. Whether there are useful classes of symbols to be distinguished remains an open question. The present point is simply that by discovering that a signal is symbolic tells one nothing about its characteristics.

Table 8-II

Semantic Properties of Optical Signals

A summary point to be made about the semantics of optical signals from the viewpoint of their characteristics is that a given signal or complex of signals can exhibit indexic, iconic and symbolic properties simultaneously. An example suffices to make the point. The male stickleback builds a tunnel-like nest and courts a female with various signals, one of which is "creeping through" the nest while the female watches. The action is indexic because it points out the nest in space and iconic because it resembles what the male's actions will be if the female deposits eggs in the nest: he will swim through the nest, depositing sperm as he goes. Other stickleback courtship actions, and perhaps aspects of creeping through itself, are purely symbolic, having no features that point out something or resemble something else.

Table 8-II summarizes the kinds of signals that are known to be indexic, iconic and symbolic. Finally, it is useful to recall that these sections on semantics concern only the classification of relations between signals and their referents, and the implications these have for the characteristics of optical signals. I have not been concerned with semantics in the dynamic sense of the communicative act, which is a broader concern and tangential to the specific aim of this volume.

coding problem

Certain relationships among signals may be governed by considerations of the communicative process itself, which I shall call syntactic problems of coding. This area has received little attention, and I consider just one example each of problems relating primarily to the sender and to the receiver.

Attneave (1954) and later H.B. Barlow (1961a,b) drew attention to the coding principle Barlow calls "message compression," in which it is reasoned that common sensory inputs should be represented neurophysiologically by brief neural events requiring little energy. Dawkins (1976) has extended this notion to motor outputs of animals, and it seems reasonable to extend it further to signal-production. Under this principle of signal-economy, commonly used signals should be brief and require little energy for production whereas longer and more costly signals should be rarer. Zipf (1935) pointed out that word-length was inversely related to frequency of occurrence, but I cannot find a quantitative example from animal optical signals. [This principle is not to be confused with Zipf's "law" or "principle of least effort," which states that log frequency of occurrence is inversely proportional to frequency-rank. Mandelbrot (1953) showed that relation to be a mathematical consequence of dividing time randomly into segments. Moles (1968) suggested that departures from this random expectation will prove to be the instructive aspects of analyzing communication, and Schleidt (1973: 367) has provided an example.] It is a common impression among ethologists, though, that frequent signals of animals are those of short duration, so the notion of signal-economy provides a useful hypothesis for future studies.

From the receiver's viewpoint there may be limits on the total number of optical signals that can be reliably discriminated, and such limits should affect the signals used by senders. At least two factors are relevant to signal-discriminability: the total number of different signals and how easily each is discriminated from the others; Moynihan (1970) and Johnston (1976) explore this topic in some detail. One may, for instance, be able to discriminate a hundred different geometrical shapes, given juxta position and adequate viewing time. However, if these shapes are seen in succession rather than simultaneously, and for short durations, confusion may ensue.

There is no limit to the amount of information that can be transferred by the presence or absence of one signal-type; for example, all real numbers can be represented in binary notation (1 and a place-holder for its absence, for which 0 is used). The disadvantage of the binary system is that a long string of numerals may be required to denote a number that can be written more parsimoniously in base-ten (e.g., 111111 in binary is simply 63 in decimal notation). Numbers could be written more parsimoniously yet with 20 or 50 different numerals, but such a large set might prove unwieldy because so many visual discriminations would be required.

Smith (1969) pointed out that signal-repertoires were often limited in size. Moynihan (1970) attempted to compile the signal-repertoires of various vertebrate species, finding fishes to have 10 to 26, birds 15 to 28 and mammals 16 to 37 displays, depending upon the species. Wilson (1972) reports that analyses by C.G. Butler and himself show social insects to have between 10 and 20 displays. These figures are based on totals for all modalities, not merely optical signals, and they are restricted to ritualized signals (ch 2). What constitutes a ritualized signal is difficult to judge objectively, and in any case these numbers probably underestimate the actual repertoires of information-transmitting behavioral patterns. For example, Moynihan's (1970) compilation of Altmann's data for the rhesus macaque yields 37 displays whereas Schleidt (1973) recognizes a repertoire of 59 "elemental key sign types" based on the same series of studies. When the behavioral units overlap in time they may be recognized by receivers as a new type, and of course the role of environmental context (ch 2) cannot be ignored. The point is that objectively defining the repertoire is difficult if not impossible. However, by using the same criteria it may be possible to compare repertoires of related species to see what factors dictate the size of signal repertoires.

One way to decrease signal-ambiguity, and hence allow for more types of signals and more parsimonious coding, is by low variability in the signals themselves. Color-patterns and morphological structures may show low variability from individual to individual, and from time to time in the same individual. Behavioral signals, on the other hand, and color-patterns that are very labile, may vary from performance to performance, as well as among different individuals of the same species, and so are especially prone to variability as signals. Ethologists have long recognized that signal-movements are more steroetyped than other behavior, but because of the technical difficulties of measuring stereotypy by means of motion picture film, often obtained under trying field conditions, quantitative studies have been slow to appear. I recall the bitter cold winds endured by Dane and Walcott during their pioneering studies of goldeneye displays filmed off the Massachusetts coast in winter (Dane et al., 1959; Dane and van der Kloot, 1964). Film studies of other avian signals (Hailman, 1967a; Wiley, 1973) and signals of other animals such as lizards (e.g., Jenssen and Hover, 1976) are now available, but it is still too early to make meaningful comparisons among species (see G.W. Barlow, 1977). Stereotypy may also be important in conspicuousness (ch 7).

In future studies of stereotypy of optical signals it will become increasingly important to distinguish the information-carrying sign-vehicle (ch 2) from the inclusive signal. Golani (1976) has shown that in what appears to be highly variable behavior in social interactions in mammals, certain aspects of the behavior are not at all variable. Although his examples deal with behavior that involves primarily chemical and tectile stimuli, the same principle may apply to optical signals, and needs to be given experimental scrutiny.

It follows that if signals are to have low variability to prevent ambiguity, they should also be as different from one another as possible. The extreme case was recognized by Darwin (1872), who pointed out that signals arising from "opposite emotions" should have "opposite" characteristics. Oppositeness is a difficult quality to define objectively, but the general notion of Darwinian antithesis is widely documented in animal signals, particularly in threat and appeasement signals (see figures 23 and 24 in Hailman, 1977a). Baylis (1974b) illustrates colorational antithesis in the fish Neetroplus nematopus, in which the social, schooling fish is gray with a black bar and the solitary, breeding territorial fish is gray with a white bar.

No discussion of economy, repertoire, variability and discriminability of signals is complete without a reminder that the environmental context of communication (ch 2) is a potentially important factor in any signaling. For ex ample, repertoire size need not be so great when differences in context are the primary information-carrying aspects of communication, and the important aspects of discriminability may be among contexts rather than among signals.

destinational problem

Certain signals or aspects of signals may determine the intended destination or addressee of other signals. Such signals are therefore semantically indexic, but also raise syntactic problems because their information relates to other signals for which they point out the addressee.

Simple approach to within signaling distance or transmission at some particular time may help determine an addressee. Here I am more concerned with particular aspects of signals, especially orientation with respect to the intended receiver. By being able to see the addressee, the sender can insure that his signals are sent in that direction and that the optical channel between him and the destination of his signals is clear. If the animal is sufficiently large, he may not be able to tell whether the addressee can or cannot see some part of his body distant from his own eyes. Therefore, one may expect encephalization of optical signals to be favored under these conditions: large animals living in environments with opaque noise (ch 7) when specific addressees are intended. It is therefore not surprising that trunk postures of elephants and facial expressions of equines are important optical signals in social interactions (Tembrock, 1968). A different factor favoring encephalization was mentioned previously in a different context. Because sense organs are directed toward objects of interest (e.g., potential predators), and sense organs themselves are encephalized, one expects to find encephalization of indexic signals in general.

A topic that follows directly from addressing the signal is monitoring its reception by the addressee. If the sender can see the addressee's eyes, then the sender knows that at least his eyes can be seen. Eye-contact is therefore a reciprocal signal in and of itself, one whose syntactical aspect relates to the signals of the other communicant. E. Burtt (pers. comm.) reminds me that eyerings of birds are important signal-markings for locating the eyes. Because the iris color of most species is very dark, birds with light head coloration have no special eyering coloration, whereas species with dark heads have contrastingly light eyerings that serve to locate the eyes.

In turn, the act of breaking off eye-contact by rotating the eyes or the entire head or body so as to look elsewhere becomes a dramatic signal indicating non-reception. These have been called cut-off signals (Chance, 1962), but they are not in all cases necessarily due to syntactic considerations. For example, "facing-away" or "head-flagging" of gulls, in which the members of a mated pair simultaneously turn their faces away from one another, also results in their hiding weapons, namely bills (Tinbergen, 1959), and hence may carry other meaning than breaking eyecontact: they demonstrate antithesis of threat.

Extrinsic signals may also be used to index the addressee. In many species of puddle-ducks the male performs a "grunt-whistle" display, in which his body rises out of the water, but his neck is arched downward with his bill in the water (e.g., Lorenz, 1941). The lateral movement of the bill sends an arc of water flying, and McKinney (1975) showed that males perform the grunt-whistle only when spatially aligned such that the water-jet moves directly toward the female. Some of the preliminary shaking movements that precede the grunt-whistle may be involved in positioning the male and aiming in preparation for displays (see Simmons and Weidmann, 1973).

Interpretational problem

Some signals provide aids to the interpretation of others by the receiver. Again, little attention has been given this subject in relation to optical signals, but at least three kinds of interpretation have been reported: behavioral context, punctuation and quantification.

A signal concerning behavioral context tells the receiver how to interpret, in a general sense, an entire group of signals. The best-documented and most discussed example of contextual syntax in optical signaling is the case of special "play signals" which G. Bateson (1956) called "metacommunication." Metacommunication is, strictly speaking, communication about communication, so that any signal with syntactic content really qualifies as metacommunication. It remains to be seen whether the broad or restrictive usage will prevail, and perhaps the term is best dropped as already ambiguous. At any rate, Bekoff (1975) has shown that coyotes may flex the forelegs, extend the hindlegs and wag the tail (fig 8-3a) before giving other optical signals such as baring the teeth, flattening the ears and erecting the mane. When the former "play" signals precede the latter "threat" signals, the receiver shows a low probability of submissive behavior, but when "threat" is given without the contextual syntax, the receiver shows a high probability of submissive behavior.

Contextual signals may also be used deceptively. In fact, such feigning is probably the chief category of intraspecific deception. For example, Carpenter (1964) notes that monkeys use sham feeding in order to get close to a conspecific for performance of behavior unrelated to feeding: as an introduction to play (p. 27) and by males as an approach to females for copulation (p. 61).

There are several nomenclatural and definitional problems concerning interpretational signals that are worth considering briefly. First, behavioral context discussed above refers to the general kind of activity being performed by the sender; environmental context discussed in ch 2 refers to information coming from sources other than the sender that helps decode the sender's signals. The latter is a more general concept, which may provide information about the sender's behavioral context or about other aspects of his signals. Second, general terms such as metacommunication (Bateson, 1956) or grammar (Schleidt, 1973) seem best used as synonyms for syntax, since it is useful to separate various kinds of syntactic signals. Lastly, there may be no firm line between signals denoting behavioral context, discussed above, and those that direct how other signals shall be read in a more structural sense.

Fig 8-3. Syntactic interpretational signals. The posture of the coyote (left) denotes the behavioral context of “play” and the tail-wagging of the mallard (right) may be a punctuational signal announcing other displays to follow.

Punctuational signals are those that specify how other signals should be read in a structural sense. For example, although tail-wagging in mallards (fig 8-3) has several non-communicative functions, it occurs with high probability just before and after sequences of pure display behavior (Hailman and Dzelzkalns, 1974). Its sequential placement therefore can transmit syntactic information similar to that of a capital letter at the beginning of a sentence and a period at the end: punctuation. Jenssen and Hover (1976) report similar small head movements of anoles given before, after or between qualitatively different segments of primary display. Such movements prior to display may also be indexic in that they attract visual attention of receivers. Schleidt (1973) discusses walking toward and walking away from receivers in rhesus macaques as signals that initiate and terminate behavioral interaction. All these punctuational signals share a characteristic identified by Bekoff (1976) in signals of play context: they are relatively frequent and of relatively short duration. The correlation between these two variables is expected from the principle of signal economy, discussed previously.

The final class of interpretative signals in optical communication is quantifying aspects or "signal intensity. It seems possible that quantification is merely a subset of a broader set of signals and aspects of signals that modify others, analogous to the way in which adjectives and adverbs modify nouns and verbs. However, quantifying aspects seem to be the only clear modifiers reported in optical signaling. The most detailed study of signal intensity is Brown's (1964) work on the Steller's jay, which raises its crest to various angles (see figure 12 in Hailman, 1977a). Brown documented quantitatively the degree of crest-raising in various social situations and found it to correlate primarily with "increasing resistance of the opponent" in agonistic encounters. This finding suggests that the degree of crest-raising quantifies something like the probability of attack by the displaying bird.

There has been much discussion among ethologists as to why quantifying signals show continuous or discrete variation, ever since D. Morris (1957) pointed out that some signals showed the latter pattern, which he called "typical intensities." Sometimes continuous variation is referred to as analog, and discrete variation as digital (e.g., G.W. Barlow and Green, 1969). J.L. Brown (1975) has criticized such usage and prefers graded and discrete. In fact, discrete variables of mathematics may be graded (show ordinal or higher level variation), and the most straightforward language seems to be simply that of mathematics: continuous and discrete (able to take on any value or able to take on only certain values along a continuum). D. Morris (1957) suggested that discrete variation or "typical intensity" was useful in standardizing a few points on a continuum that the receiver could readily distinguish (see discussion above under coding). However, no discrete system of variation has been investigated with the rigor of J.L. Brown's (1964) study of continuous variation in crest-raising, and it may turn out that discrete systems involve factors not heretofore considered.

It seems worth emphasizing a point that is obvious upon thought, but might escape notice: not all signals can be readily quantified. For example, the male cardinal is red and has no physiological mechanisms for creating degrees or shades of redness in behavioral interactions. Fishes, with their melanophoric systems that respond rapidly (ch 4), can show degrees of coloration. Many shapes, such as the crest-erection of the Steller's jay, can show degrees, and most movements can be quantified by amplitude or frequency, or both. When pragmatic needs of signaling require quantification, certain signals may therefore be evolved more readily than others. Furthermore, when two or more qualitatively different kinds of signals are to be superimposed, and each quantified in some way, there are further constraints on the choice such that the full ranges of quantification of both signals are compatible (fig 8-4).

Fig 8-4. Superimposability of signals. In canids (left) opening of the mouth and flattening of the ears may be varied independently to create all possible combinations (after Tembrock, 1968, from Eibl-Eibesfeldt). In the domestic cat (right) several signal-elements may be superimposed including flattening of the ears, rotation of the ears, dilation of the pupils opening of the mouth, etc. (modified from Leyhausen, 1956).

Some of the syntactical problems of optical signaling are summarized in table 8-III (next page), which indicates suggested solutions or expected signal-characteristics for specific problems.

Table 8-III

Syntactic Considerations of Optical Signals

a behavioral classification

It proves convenient in devising a tentative pragmatic classification for communication to follow more or less closely the implicit functional classification used by ethologists to organize animal behavior conceptually. In terms of behavior in which communication occurs, there are three broad functional categories: agonistic, reproductive and cooperative behavior. These, in turn, may be subdivided for closer scrutiny.

Agonistic behavior is a peculiarly ethological term denoting behavior that involves a contest over something, such as food or access to a female. (Its root is the Greek agon , meaning contest, from which "agony" also derives. Entomologists sometimes refer to agonism as "antagonism.") Agonistic behavior is usually viewed as a continuum that ranges from pure fighting or attack at one extreme, through ambivalent behavior and signaling, to pure flight or escape at the other extreme. The diversity of agonistic behavior is wide, and an implicit classification in much of the ethological literature recognizes: (1) personal space or individual distance, in which an animal attempts to exclude other animals from some volume of space around itself; (2) resource priority, in which an animal attempts to exclude others from the space around some discrete resources, such as a food-item or nest-site; (3) territory, in which an individual attempts to exclude others from a relatively large environmental space that may contain several resources, such as food, hiding places, nestsite,etc. ; and (4) dominance, in which an animal attempts to establish a general priority over other individuals without defending a particular spatial locus.

It is a convenient oversimplification to divide communicational signals of agonism into those of threat and appeasement. In a general sense, a threat signal has as its semantic referent conditional probabilities of overt attack. The probabilities are conditional upon the behavior of the receiver. For example, during establishment of a territorial boundary, a threat signal could indicate that the sender will probably attack if the receiver comes closer, may attack if the receiver remains, and probably will not attack if the receiver retreats. Regardless of which the receiver does, it may also reply with a signal. The original sender's threat may not have effected any spatial alteration in the relationships of the two animals, but the sender has benefited by gaining information about his opponent's own probabilities of attack. Often, a dispute may be settled by an exchange of signals without either animal risking injury or death through overt combat.

Appeasement signals appear to be given when an animal attempts to remain in some place without challenging the dominance of another animal. Appeasement signals appear to have as their referents low probabilities of overt attack. Threat seems to be more widespread, but appeasment signals are important in dominance hierarchies and in early phases of sexual behavior.

Reproductive behavior includes a very broad spectrum of activities which may be decomposed into temporally overlapping phases for purposes of discussion. In the first place, it is necessary to attract a potential mate of the same species and opposite sex, the process of mate-attraction. (In synchronously hermaphroditic species, early specification of sex is of course unnecessary.) Of potential mates, one or more animals must be selected for ultimate spawning, copulation or equivalent behavior of fertilization, and the selection process may be called mate-choice. In some species there is a protracted bond between the potential parents, and hence mate-choice in these species is often referred to as "pair-formation," "bonding," or similar terms. The third major phase is fertilization (spawning, copulation, etc.) and all the activating and synchronizing activities that lead up to it. Ironically, there appears to be no wide-spread ethological term designating these activities; "courtship" usually includes mate-attraction and mate-choice whereas "precopulatory behavior" usually refers only to actions immediately before copulation. For lack of a better term, I shall call this class simply Sexual behavior. Finally, I shall combine all activities relating to the rearing of offspring as parental care. Such activities could be subdivided into sheltering the young, feeding them, protecting them from predators,etc . However, parental care is virtually absent in many species, it is extremely diverse where it occurs, and it raises only a few general problems of signal-design, so relatively little special attention is accorded it in this discussion.

The third major functional category of social behavior that involves communication is cooperative behavior in a very general sense. Under this rubric I discuss all communication that is not patently agonistic and that is not restricted to potential mates, mates, or parents and their offspring. Three subcategories may be distinguished: social maintenance activities, such as allogrooming in mammals; coordinated group movements ; and special group activities, such as cooperative foraging and anti-predator behavior. Many of these may be elements of reproductive behavior, as when one animal warns its mate about the presence of a predator, but none of these activities is restricted to reproduction. Indeed, if current views of social behavior be correct (J.L. Brown, 1975; Wilson, 1975), many of these cooperative activities may be expected as particularly common elements of parental care. Cooperative behavior within the family is the most likely evolutionary origin of more general social behavior in extended groups of genetically related animals.

Table 8-IV (next page) summarizes the classificatory scheme used to discuss pragmatic aspects of optical signals. Each subcategory recognized could readily be further subdivided, but the present scheme is adequate for purposes of analysis.

threat

The general sematic referent of threat signals is conditional probability of overt attack. The probabilities depend upon the subsequent actions of the receiver: its spatial movements, returning signals and so forth. Threat includes long-distance signals such as those given by a songbird establishing a territory and at the other extreme signals given when animals are within physical striking distance of one another. In the former case, threat signals need to optimize conspicuousness, often in a complex environment, so that principles of ch 7 will apply. When the function of threat is to exclude only conspecifics, as in the case with territory in many songbirds, threat signals may be species-specific. However, many are not. My own notes on more than a dozen emberizines show head-forward threat to be highly similar among species (see figure 23 in Hailman, 1977a). Furthermore, many agonistic encounters are interspecific, in which case one expects convergence of signals to a common type (Cody, 1969).

Table 8-IV

A Classification of Social Behavior

The most general characteristic expected of threat signals is iconicity for attack. Most threat signals appear to have evolved from incipient movements of attack (Hailman, 1977a) and resemble fighting postures, movements and orientations. Often, the major benefit the sender gains from a threat signal is a reply from the receiver that allows the sender to judge the receiver's own probability of attack (Maynard Smith, 1974; Parker, 1974). When the sender can also effect retreat by the receiver, the benefits are greater. One therefore expects threat signals to be constructed so as to exaggerate the invincibility of the sender, and this exaggeration seems to have taken at least two forms: display of weaponry and appearance of large size.

Fig 8-5. Elements of threat-signals. The expression of the dog (a) and pecking into and pulling at grass by the herring gull (b) are iconic signals predicting fighting. The signals display the primary weaponry of the dog (teeth) and the gull (wings and bill). Threat displays in fishes may make the animal look larger, as in the lateral display of Tilapia natalensis (c, lower) compared with resting (upper). The frontal display of Cichlasoma meeki (d, right) also shows an increase in size over the normal posture (left). Dog after Darwin (1871), gull after Tinbergen (1951) and fishes after Baerends and Baerends-van Roon (1950).

Many threat signals display weapons used in fighting (e.g., N. Tinbergen, 1959; Hailman, 1977a) as part of the icon (figs 8-5a and8-5b) . For example, Miller (1975) reports that nearly 80% of threats by male walruses involve display of the tusks. Tusk-size is correlated with bodysize, but when the effect of the latter is removed statistically, there is still a tendency for males with the largest tusks to win supplanting encounters. More interesting is the fact that animals have evolved structures and color-patterns, that mimic weapons; for example, the manes and ear-markings of ungulates with horns (see figure 22 in Hailman, 1977a). The "automimetic" elements of threat display probably do not actually deceive opponents into thinking that the sender possesses extra weapons; rather, the icon of weapon-display is visually enhanced by the redundancy (ch 7).

If an animal appears larger (or closer) than it actually is, the opponent may be more intimidated because overt approach is itself iconically threatening. Ethologists repeatedly emphasize the importance of apparent size in threat displays, but experimental evidence on this point is ambiguous (Hailman, 1977a). Perhaps one of the best ostensible examples concerns threat in fishes, particularly frontal and lateral displays (Baerends and Baerends-van Roon, 1950). Spreading the operculum in the former and the fins in the latter display creates a larger image than if these components were not a part of the signal (figs 8-5c and 8-5d , previous page). It seems possible that threat displays could also be structured dynamically to give the impression of approach. If the cichlid’s opercula are slowly extended, for instance, the increasing size might resemble approach. Such looming aspects of threat signals are easily overlooked when one pays descriptive attention to the final posture rather than dynamic movement in signaling.

It is also possible that coloration may be utilized to increase apparent size. A light object against a dark background looks larger than the reversed polarity because of irradiance (ch 5). Leonardo da Vinci recognized that white clothing made persons appear larger than reality (Minnaert, 1954: 105), and it is possible that animals use the same principle. For example, one threat posture of the white-throated sparrow consists of tilting its bill upward to display the throat, which may appear larger than normal because of irradiance. Many other emberizines have less dramatic white throats, but a few species have black throats. The ruff also provides a counter-example. The aggressive resident males of this shorebird species maintain territories in the lek and have large, darkly feathered ruffs, whereas the appeasing satellite males have white ruffs (Hogan-Warburg, 1966; Rhijn, 1974), The coloration may have to do with conspicuousness in their particular habitat.

Threat signals may be very diverse. There may be different signals to indicate different conditional probabilities of attack for different circumstances. One aspect of signals that might be expected is syntactic qualification of the level of probability. As mentioned previously, J.L. Brown's (1964) study of crest-raising in the agonistic encounters of the Steller's jay suggests that such quantification is being communicated by the angle of the crest.

appeasement

There appear to be at least two distinct classes of signals given by an animal attempting to remain near another without antagonizing it: antithetical and iconic appeasements. In antithetical appeasement the sender gives signals "opposite" to threat signals: it faces away from the opponent, hides weapons, looks as small and distant as possible and so on (fig 8-6a ; see also figures 23 and 24 in Hailman, 1977a). Indeed, simply avoiding eye-contact is an important appeasement signal in canids and felids (Kleiman and Eisenberg, 1973), as well as in many primates and other animals. Such antithetical signals are often iconic for fleeing as well.

Fig 8-6. Signals of appeasement. Elephants (a) demonstrate Darwinian antithesis in that the most threatening posture (left) extends the trunk, points the tusks forward, and rotates the ears forward; antithetical elements (right) include pulling the trunk down and backward, lowering the tusks and oppressing the ears against the head. In the avocet (b) pseudosleeping in agonistic situations is an example of iconic appeasement, which also hides the principal weapon (the bill). Elephants after KUhme (1963), who separates many intermediate signals, and avocet after Makkink (1936).

Less obvious are more specific iconic appeasement signals that resemble behavior other than agonism. For example, many primate species present the anogenital region when approached by a dominant individual, and in some baboons the entire area is permanently swollen and reddened to resemble the posterior of a female in estrous (see illustrations in Wickler, 1968). Such icons are mimetic, but probably do not work by true deception in the sense of ch 6. In early evolutionary stages an appeasing sender might well have deceived the dominant individual, but it seems reasonable to conclude that the appeasement signal is now recognized as a signal, standing for a referent such as "I will behave more like an estrous female than a rival male." Similarly, the dominant male's perfunctory mounting that often follows such presenting is iconic in the same sense of resembling non-agonistic behavior. These signals are sometimes called "remotivating" signals, but there seems to be little firm evidence that the animals are "remotivated."

In general, one may expect that any non-agonistic behavioral pattern may be mimicked by an iconic appeasement signal. The classical example is pseudosleeping in fighting avocets (Makkink, 1936), which in many ways looks like sleeping and hence may signal to the opponent that the sender is not likely to attack or move from the spot (fig 8-6b, previous page). In this case, it is even clearer that the opponent is not "remotivated" to engage in some particular, irrelevant behavior.

Lorenz (1952: 186) proposed a type of appeasement signal based on showing the most vulnerable part of the body to the opponent. The signal was supposed to be symbolic for a referent such as "kill me if you wish as I can no longer resist." His idea was based on a signal of canids where one agonist turns his head away to expose his neck, and hence the jugular vein. Lorenz apparently misinterpreted the observations on wolves by Schenkel (1948,pers. comm.); in fact, Schenkel claims that the signal is given by the dominant animal, as a test to see if it has won. If the opponent resists the chance to attack the displayer's most vulnerable area, then the dominant animal realizes that the contest is over and his dominance is complete. I might add that in my college youth I put Lorenz's theory to empirical test by throwing down my snowballs when cornered, and opening my arms defenselessly. The result was a snowball in my face.

Table 8-V

Optical Characteristics of Agonistic Signals

Finally, in many agonistic encounters, especially those related to dominance hierarchies, individual recognition is important. Because agonistic signals are often encephalized for reasons discussed previously, one might expect variations in signals about the head region to be important cues to individual identity. For example, in domestic fowl, differences in the comb and wattle appear to provide information about individual identity (H.L.Marks et al., 1960).

Some of the characteristics expected in agonistic signals are summarized in table 8-V.

mate-attraction

The two major requirements for a potential mate are that it be of the same species and the opposite sex (except in synchronous hermaphrodites). Mating with another species wastes time and gametes because such matings often produce no offspring, sterile offspring, less viable offspring or offspring that are at a competitive disadvantage ecologically (see Mayr, 1963). There is therefore a strong evolutionary selection for making a conspecific mate-choice, and studies reveal consistent differences in optical signals among closely related species (e.g., Hunsaker, 1962).

The necessity for correct attraction of potential mates has often been used to explain dramatic differences in display coloration between male and female, and among males of different species. However, it appears that the explanation has been employed uncritically and overworked. Many closely related species whose breeding ranges overlap have highly similar signals, yet rarely interbreed. For example, males of the common and Barrow's goldeneye ducks differ only slightly in display markings (see figure 9 in Hailman, 1977a), and the virtually identical black-capped and Carolina chickadees differ primarily in only one type of vocalization. Conversely, the wood duck and mandarin duck have no potentially interbreeding species within their separate ranges, yet the males have highly elaborate display plumages (Dilger and Johnsgard, 1959). Sexual dimorphism is just as difficult to explain simply. Male and female black ducks, for example, are sexually monomorphic, yet the very closely related mallard shows extreme sexual dimorphism in coloration.

Sibley (1957) suggested that sympatic, closely related avian species that are polygamous and have short pair-bonds should show the greatest sexual dimorphism and species-distinctiveness in courtship signals. In such species males would be highly competitive for the females (part of the notion of sexual selection). However, Dilger and Johnsgard (1959) pointed out that monogamous species with long-term pair-bonds might suffer even more strongly from incorrect pairing. They point to species such as parrots that show elaborate species-distinctiveness, and in some cases sexual dimorphism, yet have long pair-bonds. And mallard and black ducks show the same general reproductive patterns, yet one is dimorphic and the other is not.

T.H. Hamilton (1961) suggested that it is not the length of the pair-bond, but rather the duration of courtship that correlates with sexual dimorphism. Migratory species with short breeding seasons must arrive in spring and immediately establish territories and secure mates if any offspring are to be produced. In species with rapid courtship, sexual and specific distinctiveness must be emphasized to prevent costly mistakes. Yet the mallard and black duck, to cite one example, are both migratory.

T.H. Hamilton and Barth (1962) extended the idea to seasonal dimorphism, where the male assumes female-like plumage during the non-breeding season, not only to avoid predation through concealing coloration (ch 6) , but also to reduce hostility in wintering flocks (also see Moynihan, 1960). Their evidence for the latter point is scanty and may not be correct. The male Baltimore oriole, which they cite as principally solitary on the wintering grounds, retains its bright orange plumage, whereas the male orchard oriole, which winters in social groups, is cited as molting into a dull, female-like plumage. However, hand-books for areas within the wintering range of this latter species illustrate and discuss only the bright plumage of the male ( e.g.,peterson and Chalif, 1973; de Schauensee, 1970).

Willson and von Neumann (1972) surveyed the "colorfulness" and sexual dimorphism in birds in both the Old World and New World and found certain statistically reliable differences among regions. North American temperate and South American tropical species did not differ in sexual dimorphism (39% possess it), but these avifaunas show higher incidence than either Europe (32%) or South American temperate birds (26%). "Colorfulness" had a different geographic pattern. It was highest in Neotropical birds (32%), less common in South (27%) and North (25%) American temperate avifaunas, and even less common in Europe (10%). They offer no new hypotheses to account for these differences. It might be that "colorfulness" is related to conspicuousness in the dense vegetation of the tropics (T. Johnston, pers. comm.). The requirement would be less stringent in more open temperate forests, and even less important in Europe, which has been largely deforested for a very long time.

In sum, no one has established a hypothesis that predicts with reasonable accuracy the degree of colorfulness, sexual dimorphism or specific distinctiveness in courtship plumage of birds. Fishes present the same problem, as various species are brightly colored or dull, similar to one another or strikingly different, sexually monomorphic or dimorphic,etc . Part of the problem in analyzing fishes is that many change color readily (ch 4) and detailed studies are required to describe all the colors shown by a species under different behavioral conditions. Among insects there is even another problem: UV--sensitivity. The black and white butterfly Eroessa chilensis, for example, has orange wing-tips in both sexes and thus appears monomorphic. However, Eisner et al. (1969) discovered that only the male's wing-tips reflect ultraviolet, so the species is actually sexually dimorphic.

As noted, the important aspects of initial mate-attraction that affect characteristics of optical signals are species-specificity and sex-specificity . Concerning the former, one expects only the characteristic that species should be discriminable on some basis (not necessarily optical); there is no clear expectation for particular characteristics when optical signals are used. Concerning recognition of sex, two factors in the design of optical signals may be expected: secondary sexual characteristics and sexual antithesis.

First, any features that are characteristic of one sex due to non-signal functions, such as weaponry of males, may be exaggerated optically or prominently displayed to assist sexual recognition. Such elements are called secondary sexual characteristics, and the structures and color-patterns that automimic weapons (see threat, above) may be further selected to act in mate-attraction (fig 8-7a). The other expectation is that sexes will be as "opposite" as possible in signaling: they will exhibit sexual antithesis (fig 8-7b). The difference between male and female may be virtually permanent, as in the coloration of cardinals, or virtually instantaneous. For example, Keenleyside (1972) says of the pomacentrid fish Abudefduf zonatus that the male's "pattern appears suddenly as the male begins courting and fades quickly when the female leaves the male's nest area."

Finally, behavior is organized in some animals such that a male establishes a territory from which he simultaneously attempts to exclude other males and attract females. Signals that appear to serve both functions are called advertisement , a common example being bird song with display of brightly colored plumage. Advertising signals may be expected to maximize conspicuousness(ch 7). Extrinsic signals may also be used to advertise: Linsemair (1967) reports that the sand pyramids built by ghost crabs simultaneously repulse other males and attract females.

Fig 8-7. Sex-specific signals of mate-attraction. The female three-spined stickleback (a) when ready for reproduction can be recognized by her shape, the ventrum disteneded due to accumulation of eggs. Her posture displays the shape prominently before the male. The female zebra finch (b) lacks the bright coloration of the male (c). The sexes are shown in antithetical postures: the male in a sleeked, head-forward threat (cf., fig 8-5) and the female in a more upright, fluffed posture (cf.,fig 8-6). Antithetical and iconic appeasement by females are elements of sexual behavior (see fig 8-9, below). Stickleback after Tinbergen (1951) and zebra finches after D.

mate-choice

Darwin (1871) raised a major question about reproductive behavior that has been accorded too little experimental attention: by what characteristics should an animal choose a mate in order to maximize reproductive success? There appear to be at least three major classes of criteria that would be useful in mate-choice: the genetic complement of the mate, the mate^’s parental abilities (when in volved in parental care) and the mate's fidelity (when there is to be a long pair-bond).

There has been considerable confusion as to how one animal assesses superior genes of a potential mate. The classical idea is that males reveal superior genetic complement in their competitive fighting over females. For example, courtship adornment, such as antlers and horns of ungulates, have supposedly developed as means of displaying superior aggressive abilities (fig 8-8a) . A female is supposed to select a male with well developed weapons because these reflect superior genes to be passed to her offspring. Although it may work this way in some species, in general the situation is not that simple. For instance, the most aggressive males may not make the best parents in birds because they spend so much time in agonistic activities that they neglect the needs of their off-spring and mate (Hutchinson and MacArthur, 1959).

It is not impossible that large horns and antlers in ungulates, and similar developments in males of other animals, reveal important aspects of ecological acumen in their possessors. Perhaps a male with large antlers demonstrates to the female that he is proficient in extracting critical nutrients from the enviroment. If such traits are genetically controlled, the female would do well to insure that her offspring receive the beneficial genes for good foraging. I am aware that this is not a traditional explanation of secondary sexual characteristics, but it is a hypothesis worth closer attention. To cite a related speculation, S. Robinson (pers. comm.) has found that male house sparrows bill-wipe more frequently in the presence of a female than they do in the presence of another male, or when alone. He suggests that this subtle signal may have arisen in evolution if females tended to choose mates who were proficient in feeding, and hence wiped their bills frequently.

When the male participates in parental care, these same aspects of morphological and behavioral signals may be further enhanced by at least two other mechanisms. Either the male might promote the offspring's learning how to feed optimally, or else the male may be a good provider of food for the offpsring (either by bringing food to the offpsring or leading the family to food). Thus, genetically determined ecological acumen grades into predictors of the mate's parental abilities, the second category of attributes that may be signaled in communication leading to mate-choice.

If success as a parent is correlated with age, mate-choice may be based partly on optical indicators of age (S. Witkin, pers. comm.). For example, the size of antlers and horns in ungulates is known to be a function of age in several species, thus providing another possible value of such structures as optical signals. Studies of individuals in primate groups, particularly great apes such as the gorilla and chimpanzee, show silver hairs to be age-indicators. In the human primate there is great variation in the age of graying (my mother's hair was snow-white in her teens, although I kept color a bit longer), so age-indicators may not be very accurate. In fishes and invertebrates, and probably to some extent in reptiles and amphibians as well, size alone is correlated with age and might play a role in mate-choice.

Ornithologists have long recognized that female songbirds may choose a mate more by the quality of territory he holds than by his own physical characteristics. She is optimizing her reproductive strategy, and it may be more adaptive to become a second wife on a good territory than the sole mate on a poor one (Orians, 1969). The territory is one aspect of predicting the ability of the male to be a good parent. Other aspects are commonly noted among birds. For example, grebes build a nest-like platform for copulation (Huxley, 1914). Because they cannot copulate in the water in the fashion of ducks, the platform is a functional structure in a straightforward sense, but its resemblance to nests is probably not accidental. It seems reasonable to assume that by constructing the platform, the potential parents communicate to one another something about their abilities to provide a good nest for ultimate eggs and young. Nest-like structures are important in the courtship of many birds (e.g., Verner, 1964), and potential parents often construct several nests within the territory before actual egg-laying. These have often been viewed as simple failures and abandonment, but it seems possible that in some cases they involve elaborate testing of the mate's parental abilities. The nests are extrinsic signals: iconic predictors of the ultimate nest to be built.

Many of the elaborate courtship displays of birds have been attributed to simple symbolic actions of unspecified origin. I believe that these should be scrutinized for predictors of parental abilities. For example, returning to Huxley's (1914) great crested grebes, the pair engages in an elaborate "penguin dance" in which both birds dive to the bottom and emerge with a bill full of vegetation like that used in constructions of the nest (fig 8-8b). Wilson's (1972) interpretation that "the collection and presentation of the waterweeds may have evolved from displacement nesting behavior initially produced by the conflict between hostility and sexuality" seems to me unnecessary. It is more parsimonious to assume that evolution favored grebes that could select their mates at least partially on the birds' abilities to gather nesting material successfully. Once receivers were cuing on nest-material gathering as part of mate-choice, evolution favored making the gathering visually obvious, and highly ritualized optical signals evolved. The more ethology has learned about principles of communication, the less it has had to rely on explanations such as "displacement activities" (see also Hailman, 1977a). Eibl-Eibesfeldt (1970: 127) describes taking food-like items away from Galapagos flightless cormorants, which usually present the materials to the mate at the nest-site. Without the iconic gift, the bird at the nest rejects the returning mate. More experiments need to be done to establish the importance of extrinsic signals in courtship communication.

Fig 8-8. Signals used in mate-choice. The male's rack of antlers in the moose (a, left) not only prepares him for fighting rivals but may also indicate to the female his possession of desirable genes, such as those that produce good foraging abilities. The ^“penguin dance^” of the great crested grebe (b, right) involves showing nest-like materials gathered underwater, and is thus an icon that demonstrates potential nesting abilities of the birds (after Huxley, 1914).

If both animals participate in courtship-building, the activities allow each to judge how well the other will coordinate activities so that would-be parents can work as a team in reproductive activities. Individuals of a species differ in their daily rhythmic patterns, and some kinds of courtship activities may test the abilities of the two sexes to coordinate their daily schedules in useful ways. It seems likely that courtship-feeding (Lack, 1940) is another example of how courtship activities help one mate to judge parental qualities of the other. If a female bird incubates eggs during the day, she may later have to depend upon the male to bring her food. If the male helps feed the young, that provides another reason to assess his food-bringing abilities before a final commitment to reproduction (Nisbet, 1973). I believe that animals learn a great deal more about the parental abilities of their mates through courtship activities than has been realized. These are not simple, arbitrarily symbolic actions; many are iconic predictors of parental abilities.

Finally, when the mate's presence after copulation is important to the success of the reproductive effort, mate-choice by the partner may focus on predictors of them mate’s fidelity. The female often has a larger investment in the reproductive effort, since she produces the nutritional-bearing gametes at relatively greater physiological cost than production of male sperm. Males may evolve an optimizing strategy of attempting to inseminate many females, so polygyny is to be expected more frequently than polyandry. McKinney (1975) relates the use of true deception in male green-winged teals, who assume the sleeping posture and induce their mates to sleep; then the male sneaks off to attempt copulation with some other female. The subject of mating systems is too involved to treat in this book, but one can predict that one aspect of optical signals relating to mate-choice may concern attempts of the female to test her mate's fidelity. A typical example was mentioned in connection with indexic aspects of semantics, above: female mallards attempt to incite their consort males when approached by an intruding male (fig 8-2a).

It follows from aspects of choosing an individual animal for a mate that they may also be signals for Individual recognition. In many cases, variation among individuals due to other causes may provide sufficient cues to individual recognition. In some cases, however, individual variation may be selected for its signal value, as in individual differences in display movements of the chuckwalla (Berry, 1974).

Finally, copulation itself (or functional equivalents of fertilization processes) may play an important role in mate-choice. Non-fertilizing copulations may take place in early phases of courtship, where such acts serve as iconic predictors of later sexual behavior and may also indicate the mate's fidelity. In some cases, elaborate post-copulatory displays have evolved, and although I believe these to relate primarily to mate-choice, they also function in maintenance of pair-bonds, and so are discussed most naturally in the section that follows.

sexual behavior

I have included under the term "sexual behavior" all those activities that take place between the time of mate-choice and fertilization. These activities are very diverse among animals, depending on the life-history strategies of reproduction, and hence relatively few generalizations are possible. Some primary considerations appear to be: overcoming the mate's aggressiveness, stimulating and synchronizing sexual development of the pair, and actual fertilization (copulation, spawning, etc.).

Because males either fight for possession of females or defend breeding territories in many species, overcoming male aggressiveness may be an important factor in the female's reproductive success. In some animals, the male may have to overcome aggressiveness of the female. This seems to be particularly the case with predatory species, such as the balloon flies mentioned previously, some spiders, raptorial birds, solitary carnivores and other animals. The general problem is therefore overcoming the mate's aggressiveness , and in most species the more aggressive sex will be the male.

Meyerriecks (1960) notes that the female green heron just keeps returning to the male of her choice after being driven off, and so eventually is accepted more because of doggedness than any special communication signals. Of course, this behavior may also be part of the male's testing of the fidelity of the female. In many species, however, females show special iconic appeasement, a common form being signals that resemble begging in young animals (fig 8-9). The male, in turn, often feeds the female (Lack, 1940), as mentioned in the previous section. Ornithologists have frequently interpreted the dull coloration of females as being an adaptation for concealment, especially in species where the female incubates during the day. T.H Hamilton (1961) suggested that in sexually dimorphic species such coloration also acts as appeasement, being antithetical to the bright plumage of the male (figs 8-7b and 8-7, p. 283). Therefore, both iconic appeasement andantithetical appeasement occur in sexual behavior.

Fig 8-9. Signals used in sexual behavior. The female may decrease the male's hostility by assuming antithetical appeasement postures (see fig 8-7b) or both mates may engage in behavior that hides the weapons and is appeasing, as in the laughing gull's facing-away (left, after Beer, 1975). The female may also engage in iconic appeasement (see also fig 8-6b) as in the kittiwake (right), where she assumes a begging posture like that of young birds (after Tinbergen, 1959). The begging posture accompanied by crouching that is iconic for the copulatory stance, is also a precopulatory signal in female gulls.

Behaviorally antithetical appeasement also occurs in sexual behavior. Emory (1976) reports that simple orientation toward another baboon is threatening; semantically the signal is indexic, syntactically it is addressing a specific receiver. Gelada males apparently consciously orient so that they do not face certain females, the bodyorientation being a specific form of sexual appeasement. Head-flagging between male and female gulls (fig 8-9)is similarly a cutoff signal thought to reduce agonism between mates (N. Tinbergen, 1959; see also Beer, 1975).

It is not clear that any special characteristics of signals that Stimulate and Synchronize the reproductive development of the mates are to be expected. The subject seems worthy of specific consideration. Immediate preparation for fertilization, however, is expected to involve icons of copulation itself, such as lordosis postures of females. Display of the erect penis in many primates (e.g., chimpanzee: van Lawiclc-Goodall, 1968) is obviously iconic. In some species it has been incorporated into earlier phases of reproductive behavior, where it is used in identification of sex in mate-attraction, and in some primates the penis and genital area have become brightly colored to increase conspicuousness and perhaps enhance shape (see discussion previously). In sexually monomorphic species such as the starling (Hailman, 1958) the female may mount the male as well as vice versa. It is not impossible that reverse mountings in birds without intromittant organs can lead to insemination, but it may be that these mountings are pre-copulatory iconic signals.

In many cases, one may expect indexic signals to facilitate fertilization. Thus, in the creeping-through behavior of the stickleback mentioned previously, the male shows the spawning site to the female, and in the display of the courtship paddle of certain fishes (fig 8-2c, p. 253) the male indexes the spatial position of the female for internal fertilization. A third example from fishes, that illustrated in fig 8-2d where the female mouth-brooding cichlid picks up sperm to fertilize the eggs in her mouth, emphasizes the diversity of indexic signals in fertilization.

Synchronous hermaphrodites are not common among higher animals, although in certain groups of fishes an individual may be one functional sex for an early period in life and the other sex later. If cross-fertilization is to occur in synchronous hermaphrodites, either the two individuals must release opposite gametes, or else at least one of them must release both sperm and eggs. Clark (1959) studied optical signaling of the serranid Serranellus subligarius and found that one fish from a group often began swimming in an S-curving pattern, blanched in color (losing the dark bands on the body and black spot at the base of the dorsal fin) and changed body shape. If another fish responded similarly, it was chased by the first until it returned to the banded pattern. The blanched fish appears more likely to release eggs and the banded fish sperm, but in aquaria one individual may release both gametes and self-fertilization is possible, so the situation is not simple. At any rate, it appears as if a form of sexrecognition is involved in immediate pre-spawning behavior of synchronous hermaphrodites.

Many species have elaborate post- copulatory display behavior, which has been little analyzed by ethologists. For example, the male mallard dismounts, assumes a particular bridling posture and whistles, then nod-swims in a broad arc in front of the female. It seems likely that these post-copulatory displays have evolved as part of mate-choice (above). Mallards copulate for months before nesting (Weidmann, 1956), and unless there is long-term sperm-storage, these copulations probably function more in mate-choice than in functional fertilization (McKinney, 1975). I believe the male's display is to announce to the female successful intromission, since in copulatory attempts that clearly do not result in cloacal contact, the male omits post-copulatory display (Hailman,in prep.). After bridling, the male swims in front of the female so that she can see again the individual with whom she has successfully copulated. The male may also be demonstrating his intentions of fidelity (McKinney, 1975).

parental care

Parental care involves many activities, but the primary communication occurs between the two mates (when both are present) and between parents and their offspring. Parental care is highly diverse in animals, many species lacking it entirely, so relatively few generalizations are possible. Many aspects of parental care involve more general cooperative behavior that may be shown other animals as well, and hence are treated in the subsequent sections. A few comments about special communication relating to parental care may be added here.

When there is a special site for eggs or young, the mates may have to agree upon the site. Depending upon the nature of the site, which is very diverse among species, special indexical signals may be involved. For example, the dorsal display coloration of hole-nesting birds was mentioned previously (von Haartman, 1957). When caring for eggs or young, the parents may have to synchronize their activities, and in some cases of birds, special "nest-relief" ceremonies have evolved, although their optical properties have not been well analyzed. Many may be simple icons for incubation and agonistic behavior (Beer, 1961; see also Armstrong, 1947).

Feeding is a primary interaction between parents and offspring, with the young animals evolving special signals to indicate that they are hungry. For example, most young songbirds gape to their parents, and the gapes are decorated with patterns of color that enhance conspicuousness and in many cases may be species-specific. Friedmann (1960) pointed out that parasitic weaverbirds, which lay their eggs in nests of other species, have young with gape-signals resembling those of the host species. In some cases it has been suggested that gape signals are icons resembling other things, but the whole area of investigation requires further work. For example, Hingston (1933) suggested that the rows of white conical projections in the gapes of nestling bearded tits were mimics of palatal teeth of reptiles to frighten off nest predators.

The parents, in turn, may have special signals concerning feeding of the young. In gulls, the conspicuous red markings on the beak serve to test the hunger state of the chicks, which beg by pecking if they require food. The parent responds to the pecking by regurgitating partially digested food, and then indexes its location by pointing to the food with its bill. Chicks pecking at the bill-tip strike the food for the first time, and thus begin to learn its visual characteristics (Hailman, 1967a). In mammals, the female may signal willingness to nurse offspring by very simple visual signals, such as lying on her side to expose her teats or straddling her young. Such behavior is iconic and in some cases indexic, and no doubt the coloration of teats in some mammals serves primarily an indexic function in guiding the young.

Table 8-VI summarizes some of the attributes of signaling in reproductive behavior discussed above. It is intended to be suggestive of the kinds of variables that should be scrutinized in order to understand optical signals, rather than an inclusive classification of all relevant variables.

Table 8-VI

Characteristics of Reproductive Signals

social maintenance activities

It is easy to see how one animal may derive benefit from allo-maintenance activities by another. For example, it is physically impossible for a bird to preen its headfeathers with its bill. The bird may be able to effect the same result by other means, such as scratching with its toes, but one solution is to have another bird preen its head. Or, if grooming in monkeys serves partly to find and remove ectoparasites in the fur, a monkey that cannot see its own back or reach it conveniently benefits from another monkey's allogrooming.

*See table 8-V; #See table 8-VII.

Social maintenance activities are not, however, always that straightforward. In many cases they have become incorporated into agonistic and reproductive behavior: subordinates grooming dominants, mates grooming oneanother, parents grooming their offspring, etc. (see Harrison, 1965; Sparks, 1965). Indeed, the behavior may cease entirely to serve as functional maintenance activities and become instead signals of various kinds. It is not my intent here to discuss these intricacies of social maintenance activities, but rather to ask what expectations one may hold about the optical signals that accompany them.

One expectation is simply that of indexing, usually by orientation, the area the sender wishes to have groomed and addressing the invitation to a particular individual (fig 8-10). There are often iconic signals of maintenance itself, such as ruffling the feathers, which is done prior to some kinds of self-preening in birds. Such signals have also evolved into deceptive mechanisms of interspecific communication, as when the brown-headed cowbird female (which lays her eggs in the nests of other species) solicits allopreening from the host birds by displaying ruffled headfeathers (Selander and La Rue, 1961).

coordinated group movement

Coordinated movements of whole groups of animals are common. Movements may be short, as in daily troop wanderings of baboons, or lengthy, as in transcontinental migrations of some birds. They may be periodic or virtually constant, as in the schooling of many fishes. The benefits of group movement may relate to specific activities, such as cooperative foraging or predator-defense (discussed below), but whatever the benefits there must be signals of some sort that coordinate and integrate the movements of the component animals. In most cases the signals will be optical (but see Pitcher et al. , 1976).

Species-specificity is not a universal trait of signals that coordinate group movements because mixed-species groups are well known (e.g., Moynihan, 1968). In such cases, there may be convergence of coloration or other attributes that promote recognition, as in the black coloration of huge interspecific flocks of birds during the winter in southern United States. (Some groups consist of literally millions of red-winged blackbirds, rusty blackbirds, common grackles, brown-headed cowbirds, starlings and other species.)

One trait expected in signals that promote group movements is iconicity for movement itself. Incipient movements of locomotion are common in birds (Daanje, 1950) and these are ritualized to various degrees for use as signals promoting synchronous action (see Andrew, 1956 for a discussion of avian tail-flicking as a typical example). In other cases, the iconic nature of locomotion-promoting signals is not immediately obvious. For example, the rapid, jerky uptilting of the mallard^?s bill, which one of my students dubbed "lid-flipping," can be seen upon scrutiny to resemble the head movement that is part of take-off from the water. However, lid-flipping is not given invariably before take-off nor do mallards always take off after lid-flipping. The action is a deliberate signal to recruit social movement: if other birds begin lid-flipping, take-off of the entire group is probable, whereas if no other birds show the action, the sender usually ceases lid-flipping and remains on the surface with the rest of the group.

Signals that coordinate movements of groups may be expected to show two kinds of indexical qualities. The first is indication of the intended direction of movement, and such signals may be very simple. For example, Kummer (1968) reports that Hamadryas baboons move off together after the night's sleep. The males face and start out in some direction, each with his harem, in an apparent attempt to recruit others. After several false starts, eventually the entire group moves in the direction indexed by a particular male. Of course, this simple signal is also iconic because it reveals the intention of the male to move as well as indexing the direction of intended movement.

The other kind of indexical signal integrates the spatial organization of the moving group. For example, Keenleyside (1955) cites the black mark on the dorsal fin of Pristella riddlei as promoting schooling in these fish (fig 8-10, next page), and Shaw (e.g., 1970) has investigated schooling responses experimentally with a striped drum such as used in flicker studies (ch 5).

Fig 8-10. Indexic signals used in cooperative behavior. At left, the invitation for allogrooming behavior of the lesser kudu (after Walther, 1964) and at right the black fin-markings of Pristella riddlei that promote schooling (after Keenleyside, 1955).

Fishes such as the alewife, menhaden and gizzard shad have black marks just behind the gill covers , whereas others such as the red drum have a spot on the caudal peduncle. Many large tuna have a conspicuously dark pectoral fin (unusual among fishes, which often have transparent pectoral fins). If these various markings are all indexic signals that help a fish position itself with respect to companions, the body locus of such signals should be investigated in relation to the geometry of schools.

The markings and geometry of schools may depend upon environmental circumstances. In reviewing the behavior of cichlids, G.W. Barlow (1974: 19) remarks that ״In general, the more the fish finds itself in open water, often schooling, the greater the tendency to develop the row of spots into a stripe. While up in the water but over rocks, particularly hovering in groups, the general pattern is for the appearance of the mid-body and base-of-tail spots. When the fish move closer to the bottom, often passing in and out among holes or submerged tree branches, one sees a combination of spots and bars with softly developed edges.״

Baylis (1974a) provides photographs of the cichlid fish Neetroplus nematopus , which may be gray with a prominent mid-body bar that is either black or white. The black bar is shown in schooling whereas the white bar is shown by parents guarding young, when they repulse all other adults (G.W. Barlow, 1974: 21 and Baylis, pers. comm.). This example suggests Darwinian antithesis in social Vs. nonsocial behavior, a type of antithesis apparently not heretofore recognized.

special group activities

Sometimes animals cooperate toward some collective end that requires for its success special kinds of coordination among individuals _י as in cooperative hunting by prides of lions or cooperative anti-predator reactions in flocks of birds. Such specific coordination is more likely to be peculiar to the species than are the simple group movements discussed above. There are, of course, heterospecific foraging groups (e.g., Moynihan, 1962; Morse, 1970), but these rarely involve a sophisticated degree of cooperation among component individuals. The success of a special group activity may require specific behavioral tactics, as in special aerial tactics in bird flocks approached by a raptor (e.g., Hailman, 1959a). In such cases, Species-Specificity of signals may be expected, an example of which is the wing speculum patterns of puddle ducks (see figure 20 in Hailman, 1977a), which help assemble homospecific groups in flight during emergency situations.

When special activities revolve around some specific place in the environment, such as a predator or food supply, one may expect the use of indexic signals to point out the specific places or coordinate special movements with respect to the places. For example, the direction of gaze by an animal giving an alarm call may index the position of the source of alarm (see fig 8-1a) , or the direction of movement of a fleeing white-tailed deer may become conspicuous by the white tail, which indexes simultaneously the ρrobable direction of disturbance as well as the intended direction of fleeing. Of course, it is possible that animals give "false alarms" to deceive their companions and thereby gain some benefit (Charnov and Krebs, 1975).

Some characteristics of signals in cooperative behavior are summarized in table 8-VII (next page), which, like preceding tables in this chapter is meant to suggest kinds of relevant variables, rather than exhaustively review animal signals.

a final note on pragmatics

It would certainly be destructive to my aims if this brief discussion on pragmatics and the classification of social behavior used to structure it were taken by any reader as a view of how social behavior and communication are organized in animals. Throughout this discussion, real social problems of behavior have been crushingly simplified in order to extract some main features of optical signals from their broad pragmatic contexts. A huge literature in ethology could be cited to allay any false impressions, but let me cited just three recent papers that might be consulted to appreciate the complexities of social interactions and use of optical signals therein.

Table 8-VII

Optical Characteristics of Cooperative Signals

Simpson (1973) provides a useful analysis of how recognition of individual animals changes the meaning of displays as well as the contexts in which they are given. Beer (1975) makes a sensitive analysis of signals by gulls, reviewing in some cases how analytical explanations have changed through the studies of G.K. Noble, N. Tinbergen and others to the present day. McKinney (1975) performs a similar service in reviewing the complicated factors that dictate the use of optical signals in ducks, which also have a long history of analysis from 0. Heinroth, through K. Lorenz and U. Weidmann and many others. These kinds of studies are pioneering the way to an understanding of pragmatics in optical communication. My aim in this section has been merely to show that optical properties of signals may be dictated by their pragmatic use, and, conversely by implication, that the study of optical principles in signaling may help to unravel the complexities of communication events in social animals.