The concluding question and answer of Whiten & Byrne's (W&B's)(1988b) valuable survey is worth repeating, if only to save W&B from a likely misreading: "But can anecdotes ever be more than a jumping off point for more systematic work? We propose that the answer must be 'no'...." I have been dismayed to learn that my own limited defense (Dennett 1983) of the tactic of provoking "anecdotes"--generating single instances of otherwise highly improbable behaviors under controlled circumstances--has been misinterpreted by some enthusiasts as giving them license to replace tedious experimentation with the gathering of anecdotes. But as W&B stress, anecdotes are a prelude, not a substitute, for systematic observation and controlled experiments.
They describe in outline the further courses of experimental work that would shed light on the phenomena of tactical deception, and suggest that the postponed question of whether these behaviors arise as a result of "creative intelligence" may be settled by the outcome of such research. This research should certainly be pursued, and if properly conducted its results are bound to shed light on these issues, but it is worth noting in advance that no matter how clean the data are, and indeed no matter how uniform they are, there is a systematic instability in the phenomenon of creatively intelligent tactical deception (if it exists!) that will tend to frustrate efforts of interpretation.
To see this, consider the range of possibilities available in the generic case, stripped to its essentials. Suppose AGENT intelligently creates a deceptive tactic that is devastatingly effective on a first trial against TARGET. Will it tend to be repeated in similar circumstances? Yes; ex hypothesi it was intelligently created rather than a result of blind luck or sheer coincidence, so AGENT can be supposed to recognize and appreciate the effect achieved. But then there are two possible outcomes to such repetition: Either it will provoke countermeasures from TARGET (who is no dummy, and can be fooled once or twice, but will eventually catch on), or it won't (TARGET is not so smart after all). If it doesn't, then the exploitative behavior will become (and be seen to be) stereotyped, and ipso facto will be interpretable not as a sign of creative intelligence but as a useful habit whose very cleverness is diminished by our lower regard for the creative intelligence of the TARGET. If, on the other hand, TARGET attempts countermeasures, then either they will work or they won't. If they don't work, TARGET will be seen once again to be an unworthy opponent, and the deceptive behavior a mere good habit. If the countermeasures tend to work, then either AGENT notices that they do, and thereupon revises his schemes, or not. If not, then AGENT'S intelligence will be put into question, whereas if AGENT does come up with a suitable revision then he will tend not to repeat the behavior with which we began, but rather some relatively novel successor behavior.
(The whole paper is now available in Daniel Dennett, Brainchildren, Essays on Designing Minds, MIT Press and Penguin, 1998.)