[psychology] the art of quick chicken sexing in iraq

This is a de-mystification of Richard Horsey's wordy article:

Most sexers just look at the rear end of a chick, and ‘see’ that it is either male or female, like expert chess players who can just ‘see’ what the next move should be; similar claims have been made for wine tasting and medical diagnosis. One thing for sure - these skills are hard-earned and not accessible to introspection. Mr. Horsey says it’s done on the basis of very subtle perceptual cues.

We are all constantly categorizing quickly and subconsciously, and the process is completely inaccessible to introspection. The question is, why is it so difficult to acquire skills such as chicken sexing, when we automatically acquire the ability to categorize other objects. In cases such as chicken sexing, where we do not automatically develop categorization abilities, our inbuilt attention directors do not cause us to attend to diagnostic cues, and our attention therefore has to be drawn to these cues in another way, such as through training.

It’s a little-known fact that the world’s best chicken sexers come almost exclusively from Japan. Poultry owners once had to wait until chicks were five to six weeks old before differentiating male from female (the sex became visible when adult feathers started appearing, on the basis of which cockerels and pullets could easily be distinguished).

But for commercial egg producers it’s important to identify the females as soon as possible, to avoid unnecessary feeding of unproductive male chicks. Enter the Zen-Nippon Chick Sexing School, which began two-year courses training people to accurately discriminate the sex of day-old chicks on the basis of very subtle cues.

If you ask the expert chicken sexers themselves, they’ll tell you that in many cases they have no idea how they make their decisions. In some cases, we learn to discriminate objects on the basis of casual exposure to instances - dogs, cats, our friends and relations, chairs and water. In other cases, no amount of exposure seems to do the trick, e.g. poultry.

Consider the case of soldiers who must learn to distinguish friend from foe in quickly and accurately, and must often be based on a degraded stimulus (a tank glimpsed briefly through trees; a fast-moving aircraft seen in the distance). The task is made more difficult by the fact that a large number of sometimes very similar stimuli need to be discriminated. Subjects do develop some ability to discriminate but these skills can be significantly improved through training.

The need for rapid aircraft identification first became vital during the second world war. Expert ‘spotters’ did exist, but there were too few of them. Training centres were therefore set up, but the problem was that the experts had no idea how they had acquired their skills in the first place, or how to transmit those skills to others. Training regimens therefore had to be developed somewhat by trial-and-error.

Special-purpose mechanisms which evolved for similar tasks (face recognition, animal kind classification, and so on) are not of much help in these cases. To improve speed and accuracy, current training involves attending to certain diagnostic cues.

Bird-watchers need to make quick and accurate identifications on the basis of often impoverished information. While some ability to does develop from casual exposure, explicit instruction is required, often on the basis of bird books, which point out the features that are diagnostic of particular species.

One feature of expert chicken sexing that has attracted the interest of philosophers (e.g. Brandom 1998) and psychologists (e.g. Harnad 1996) is that the chicken sexers report that in many cases they do not know on what basis they make their discriminations:

“To be close to 100 per cent accurate at 800 to 1200 chickens per hour for a long day, intuition comes in to play in many of your decisions, even if you are not consciously aware of it. As one of my former colleagues said to me…‘There was nothing there but I knew it was a cockerel’. This was intuition at work.” This is interesting to philosophers because it appears to be a case of knowledge in the absence of (rational) justification.

It is interesting to psychologists because it is a capacity which, although learned, appears not to be accessible to introspection. A similar feature has been reported with aircraft identification, as mentioned above.

After a great deal of practice, many bird-watchers can identify a bird by the “way it looks”, the bird’s “jizz” and birdwatchers perceive the jizz as a gestalt, but cannot say what the features are that make up the whole.


Our detection abilities are also fast and accurate, and usually a briefly presented or partially occluded object can still be identified; the cluster of properties that have been reported for expert detectors in the various domains we have looked at. The interesting question is why it is so difficult to acquire these skills.

Berretty et al. 1999 within Gerd Gigerenzer’s “fast and frugal heuristics” framework, postulates simple rules that allow us to make accurate decisions under time pressure. They give this example: First, we are invited to consider the hiker who comes across a bird while walking in the alps. In trying to identify the bird, the hiker takes out the bird book and uses a number of features of the cues to identify it. Second, we are asked to consider the rabbit, who has a far more limited aim when it sees a bird: identify it as predator or non-predator, as quickly as possible and get away.

Since its survival is at stake, the rabbit does not have the luxury of the hiker’s strategy. Berretty et al. suggest “categorization by elimination”. Cues are accessed sequentially in a pre-determined order, and each cue eliminates candidates from the set of possible categories for an object. In the case where all cues are exhausted, and more than one possible category remains, a random assignment is made.

Cue order has a large impact and it is therefore vital for organisms to be able to consider cues in the most effective order. Berretty et al. propose that cues may be ordered by determining their effectiveness when used alone. More effective cues are then ranked higher than less effective cues. This implies that the same cue order is used for all objects in a particular domain.

There are several ways in which cues and cue orders can be acquired. Gigerenzer & Goldstein (1999: 92) suggest genetic coding, cultural transmission, and direct observation. However, in the cases already mentioned, expert categorization seemed to be a gestalt phenomenon, rather than being based on individual features. It may be that expert categorization is a truly gestalt phenomenon which cannot be broken down or it may be that expert categorization does take place as a result of the recognition of specific features, but that since the process is not accessible to introspection, we are not aware that categorization is feature-based.

The evidence seems to point to the latter possibility.

It is not wildly implausible to suppose that we might also be able to detect some animal kinds in certain situations on the basis of a single feature, general shape. This might also qualify as a gestalt process. But it seems unlikely that any gestalt process could be involved in categorization of fast-moving aircraft seen in the distance, briefly glimpsed birds, or complex configurations of chick genitalia.

Consider visual word recognition (Rayner et al. 2002). The “whole-word” approach to the teaching of reading uses rote to recognise at a glance a basic vocabulary of words. They then gradually acquire new words through seeing them used in the context of a story. This is in contrast to phonics, which explicitly teaches the connections between letters and phonemes (including exceptions).

The moral seems to be that even automatic categorization processes that appear to be based on gestalt properties are in fact based on discrete cues. So how does one actually become an expert chicken sexer: acquire new cues, or acquire a new ordering for your existing cues? This is not enough by itself.

Evolution could give rise to the genetic coding of cues or a predisposition to learn cues. We might be innately disposed to use specific cues; we may also be endowed with mechanisms that allow us to acquire further cues. Cultural transmission can also play a role, through apprenticeship, education, or other forms of knowledge transfer. Lastly, we could also learn cues by direct observation.

Cues acquired through cultural transmission or direct observation are presumably accessible to consciousness. This suggests that the cues used by experts when they make fast, unconscious categorizations must be those that are acquired via subconscious and presumably innate mechanisms.

If this is correct, then it would seem to imply that learning by cultural transmission or by direct observation is not necessary. But in fact these kinds of learning do appear to be necessary: people do not spontaneously become expert chicken sexers. There must be other factors.

The first is ‘causal cognition’ (Sperber et al. 1995), i.e. if the first event occurs again in the future, it is assumed that the second will also occur. The second is implicit learning (Lewicki et al. 1992; Cleeremans et al. 2000), where we are not aware of what our subconscious is doing.

All of this explains why, even though the cue acquisition processes are automatic and subconscious, it takes a great deal of time, effort, and often explicit training for people to develop expert categorization abilities in many domains.

References and fuller psychological explanation on Richard Horsey’s page.
http://www.phon.ucl.ac.uk/home/richardh/chicken.htm