The Logical Analysis of Animal Communication

1. What they did

Animals clearly communicate, but in 1961 most attempts to study their signals started from human language and asked what animals lack. Marler argues for a strictly objective approach: watch the response a signal evokes in other animals, and from that response infer what information the signal carries. Only an outside observer can do this, by recording what the receiver actually does.

He borrows a framework built for human language by the logician C. W. Morris (1946), which sorts signs into four types:

• Identifiors (Marler suggests “locators”), which tell the receiver where and when to direct a response.
• Designators, which tell the receiver about the characteristics of an object, such as an appropriate mate.
• Appraisors, which help the receiver choose between options.
• Prescriptors, which tell the receiver what to do.

He tests the framework on examples from the published literature, especially the song of the male chaffinch and the dances of honey bees, and then asks how the kind of information a signal carries shapes its physical form and its evolution.

2. What they found

The chaffinch song as a worked example

The male’s song tells an unmated female in breeding condition where he is (locating information), that he is a chaffinch, a male, in breeding condition, holding a territory and without a mate (designative information), and prescribes that she approach him. How often he repeats the song probably helps her choose between males (appraisive information).

Five kinds of designative information

Animal signals proved rich in designative information, which Marler splits into five kinds:

• Species-specific: the signal tells the receiver the sender is one of their own species. Such signals tend to be distinct from those of other species, stereotyped and conspicuous. Some signals lack it: the chaffinch’s thin, high hawk alarm call is shared by several small woodland birds, and chaffinches respond to the other species’ calls as promptly as to their own.
• Sexual: the signal tells the receiver the sender’s sex. How much the signals of males and females differ varies greatly between species.
• Individual: the signal identifies a particular animal, which matters for recognising mates, rivals, young and companions. This needs signals that vary little within an individual but differ between individuals. Some bird songs seem to carry both species and individual information in different parts of the song.
• Motivational: the signal tells the receiver what the sender is likely to do next, such as attack, flee, feed or mate. Some signals are all-or-none (the “principle of typical intensity”); others grade finely with the sender’s mood, as in fighting displays and in the cat facial expressions drawn by Leyhausen (1956), which change with the balance of aggression and fear. A second group of “antithetic” signals, like the posture of a submissive dog, carries negative information: that the sender will not attack or flee.
• Environmental: the signal tells the receiver about the world around the sender, mainly food and predators. Examples include the herring gull’s food call, the hen calling chicks to grain, and the honey bee dances, which give the direction, distance and richness of food or of a new nest site. Most danger signals only grade the level of danger rather than naming the predator. The chaffinch gives the hawk alarm call only for a flying hawk in early spring, but later, with nestlings, for many animals near the nest. The willow warbler is thought to have two mobbing calls, one for a perched hawk and one for a cuckoo.

Context matters

The same signal can mean different things to different receivers. A male chaffinch intruding on a territory flees on hearing the owner’s song, which tells him the owner is ready to attack. Senders may also give several cues at once: in rats, smell, touch, sight and sound may all play a part in courtship.

How signals are built

• Information is usually carried by the whole signal, not by separate parts that can be recombined as words are in human language. Because of this, what the signal designates and what it prescribes usually cannot be separated.
• Signals that vary continuously suit appraisive information, subtle shifts in mood, and environmental information that varies continuously (such as the bee’s direction and distance). Discrete, stereotyped signals make up most of the chaffinch’s calls and are quicker and easier to interpret.
• Many signals are not arbitrary. Calls used when mobbing an owl are short, repeated clicks that are easy to locate and draw others to the owl, while the hawk alarm call has a structure that makes the caller hard to locate. Threat sounds may echo the snapping of beaks in a fight, and many visual threats grew out of “intention movements” such as baring the teeth.

3. What they concluded

• Judging signals by the responses they evoke gives an objective way to study animal communication without importing assumptions from human language.
• Morris’s framework transfers to animals, but needs extra divisions: designative information is the richest category in animal signals and falls into five kinds.
• The type of information a signal carries shapes how natural selection acts on its form, for example stereotyped signals for species identity and graded signals for subtle changes in mood.
• The usual split of signals into “arbitrary” and “iconic” is too crude. Signal structure can relate in different ways to each kind of information it carries.
• The limited flexibility of most animal signals is not necessarily a lack of brain power: life is dangerous, so signals must be quick and easy to read, often without practice. The most elaborate known system, the honey bee dance, comes from a highly organised society.
• A full account of signal evolution would also need comparative data, the genetics and learning of signals, and how animals learn to respond to them, which is beyond the scope of this essay.

6. Other technologies, skills or innovations

• Applies semiotics, the theory of signs from logic and linguistics, to animal behaviour, at a time when information theory (Shannon and Weaver) dealt only with the physical form of signals and not their meaning.