Debating about Animal Signals & Communication
Letters to John Maynard Smith: August 21, 2000
The following is my second letter to John Maynard Smith.
21 August, 2000
Thanks for your letter. This back and forth dialogue is important, at least to me, to make it clear where we stand. I wish it happened more often among scientists in this field. I must admit that your sentence “my difficulty arises because I cannot believe you really mean what you appear to be saying” worries me, mainly because I am not sure what is it that I appear to be saying. On the other hand, your scenario of roaring in stags shows me we do agree on mechanisms and signals, not necessarily on terms.
John, I think I have managed to pinpoint our differences in view and definitions regarding amplifiers and indices. It looks like stemming from an attempt to take a current definition (index, from semiotic theory), and adapt it to biological signals. As you can see below, I have tried to reconcile the two as much as I could, with only little success, and I am afraid even that did not work in the line of your previous thoughts.
First, the way you described the evolutionary scenario of roaring in stags, I would agree that low-larynx would be an “index” in my terminology, and it is, indeed, parallel to the arched back of a threatening cat. It is not an amplifier in any rate. In part, and forgive me for saying so, I can blame you for my decision, in my 1997 paper, to use the term “index” to account for such signals. An easy way out, isn’t it? But seriously, in your paper with Harper you used “index”, in my opinion, to denote both amplifiers, and signals such as scratches by the tiger. I felt it was necessary to discriminate between these two, and doing so I apparently used it the opposite of your intentions, at least the way you have expressed them here, as you seem to prefer to replace “amplifier” for “index”. I don’t, and my reasons, as I explain below, are not purely selfish.
Just a side kick here: I have chosen to use the term amplifier because such signals increases amplitude, not intensity: amplitude of perceived differences, but also, and mainly, amplitude in the corresponding response of recipients (giving greater benefits to high quality, greater costs to poor quality). If I may use Hasson 1994, 1997’s terminology, then by producing the signal, the effect of signal recipients on the signaller’s S-component is amplified (increased in amplitude).
I agree that there might be a difficulty with my use of the term “index”, if you want to make it consistent with the semiotics’ use of this term. If you want to find a different, new term for such signals, I don’t think I would object, as long as it is not another loaded term. In your dictionary, if you make one, you could make a reference to “Hasson:index”.
In any rate, I think the main source of confusion lies in your paragraph where you define index. The first sentence that alarmed me was “But I cannot accept the use of “index” to mean an act that decouples the level of a signal (pitch) from the quality signalled (size)”. Of course you can. Your scenario of roaring had already done it. The pitch is the cue used to determine size, not the signal. The amplifier (primary signal) is the making of exhaling sound during an effort louder by roaring, and the secondary signal (Hasson:index, or whatever term you choose for it) is lowering the larynx to further lower the pitch, as you have explained it. Here, there might even exist a third level signal, if roaring is intensified to the point it becomes difficult to produce, and then roaring rate may become a cue that, when intensified further till (i.e. the rate), becomes, at equilibrium, a handicap (higher rate for bigger, stronger stags). It seems to me that the pitch itself remains at all times only a cue used to determine size, and is never a signal.
Then you define index, according to semiotic theory, as “a signal that cannot be faked because its intensity is physically connected to the quality being signalled.” My first thought was that this definition may fit both amplifiers and Hasson:index. But it does not. It fits neither. The problem again, I am afraid, is a failure to discriminate between the signal and the original cue.
I think a good hypothetical example is the following: the more erect an animal stands during threat or courtship, the taller it would appear. Is that an index? At first sight you might say: yes. This is a signal that cannot be faked because “its intensity [tallest point] is physically connected to the quality being signalled [size].” However, the signal is “standing erect”, not size. Size remains the original cue used to assess the signaller. Standing erect only makes the signaller appear taller. Hence it is not the intensity of the signal that is physically connected to the quality being signalled. It is the tallest point in the body, usually the head, that is physically connected with size. I don’t remember now how exactly an index is defined in semiotics (and, being away from the academic life now, I don’t have the time to look for it, sorry), but if this is it, it is not useful to signalling. It is not a pointer, in any rate, whereas both amplifiers and Hasson:index are.
I can think of two signal types that may correspond, perhaps, to your definition of an index. (1) Attention signals (informing of presence of signallers): the louder or the more conspicuous the signal, the more likely it is, from the recipient point of view, that the signaller is present (greater probability of being heard or seen). (2) Some handicaps. For example, if size is important, and recipients respond to size, then increasing size would be a signal (i.e., that component of size that evolves as a change in the S-component of fitness only). At equilibrium, this would be a handicap (trade-off between cost and benefit, and different optima for different individuals, based on their quality). Nevertheless, such a handicap is also an index, because its intensity (the signalling component of size) is physically connected to the quality being signalled (size).
The use of the term “index” is not necessary in the first case, and may have an explanatory value only in the second.
Note again that both an amplifier and Hasson:index are pointers. An amplifier’s intensity improves perception of the cue or signal (e.g. contrast of an outline of a feather or fin improves perception of damage, or increased sound [such as roar] improves perception of the pitch produced during exhaling while making a physical effort). Hasson:index changes perception of the cue. The apparent intensity of the cue (size) changes with the intensity of Hasson:index (standing more and more erect, or lowering the larynx further and further). Seeing it this way then, still, Hasson:index is nearer to the semiotic’s index than amplifiers, is it not? It is not too unlike the handicap-index, by the way, though the mechanism is different.
One last comment regarding signal design and amplifiers: an outline improves perception of damage to the margins of a feather or fin, though its intensity (contrast) has nothing to do with damage (except that melanin may diminish damage). Hence, there is a special design of a line (i.e. on the margin) that gives this information away in the best possible way. Roaring would best give pitch away. Hence, different cues (or signals) can be revealed best by a special design that makes quality impossible to fake (within the constraints you mentioned). Nevertheless, this is NOT an index as you defined it. Amplifiers are pointers, and quality and signals ARE decoupled. It is my view that most handicaps should also have special designs that make them specialized on handicapping particular qualities (Hasson 1997), but this is a different issue.
“What to do?”
At first I only wanted to say: well, I have used the term amplifier since 1989, and this term has already been established, why confusing people? Moreover, I was the first to describe under what condition amplifiers may evolve (Hasson 1989, Hasson et al 1992), and courtesy might be in place (remember, for amplifiers poor quality signallers lose by revealing their poor quality, hence a model was necessary to explain how it works; this is not necessary in the case of Hasson:index). Moreover, there is a “positive” amplifier and a “negative” one (attenuator). As evolutionarily paired signals, amplifier-attenuator is a more appropriate pair of terms than index-attenuator. So these alone are several good reasons why not abandoning the term amplifier.
Now, however, seeing more clearly the difficulties with your original “index”, my objection to replace the term amplifier with index has a more robust foundation than the reason of mere precedence and courtesy. I hope you would agree.
Definition of signals and knowledge:
Regarding a definition of signals and “knowledge”, my previous post-doc, Phil Taylor, now in Cincinnati, has recently told me he was concerned about this too. Apparently an important point in the way I use the term “decision” did not come across well. When I am talking about the influence of “knowledge” on decision making, I do not mean it to necessarily be a conscious decision. Sensory stimulation would qualify, in my view, as a change in knowledge.
Changes in behavior in response to stimuli mean that an organism responds to changes in perceived variables in the environment. A Stimulus implies that something in the environment is not fixed (otherwise there would be no change in behavior), and that organisms respond to different values of this variable. Responses could be evaluated or be automatic (such as responses to pure attractors [e.g. Fisherian signals], to handicaps or to many cues). Knowledge, at least the way I meant to use it, includes values assigned to all variables, regardless of the type of response, as long as there is one. Therefore, a perception of a change in day length is a change in knowledge, even if it is made by a plant, as long as it normally affects its behavior. It does not matter, in my opinion, whether a “decision” is made through the neural system or by another physiological mechanism. Environmental factors that do not elicit response, hence cannot be defined as stimuli, are not part of the knowledge space, and cannot affect decisions.
Perhaps this was not clear enough in my paper, but this is how I mean it, and again, I don't think we should really disagree here.