Monday, September 15, 2008

With Specimen in Scoring Position

Inside 'Dry Storeroom Number 1,' in the basement of London's Natural History Museum, is the “type” specimen of the sunfish.

According to the linked above, "a type specimen is the official example of a given species, against which all creatures like it can be compared."

It is important to note that a type specimen is not necessarily typical, or average, but archetypal. An exemplar, the most blankiest instance of any given blank.

Often after a great feat, a ballplayer sends his spikes, or glove, or the ball- something commemorative of the moment- to the Hall of Fame.

Tied at 5 Saturday night, in the 8th inning, with Lowrie on third as the go-ahead run in a game the Sox once trailed 5-2, Jacoby Ellsbury, taking a full swing on a Scott Downs delivery, meekly tapped the ball about 30 feet towards first base, nestling just inside the line. Downs, in his rush to throw out the speedy Ellsbury, slipped, sprawling on his chest. The ball, with little resistance, came to a rest, just inches fair, and Lowrie scored what would be the deciding run.

Of course, they should send that ball to Dry Storeroom Number 1. It's not a typical Ellsbury hit, but it's an archetype, an exemplar, the official example of an Ellsbury cheap shot, helped along its slow slow path by the threat of speed. As such, it is the example against which all others are to be compared.

Towards that end, in Sunday's game, Ellsbury came to bat in the 2nd inning against Halladay, with another runner on third, and again with 2 out. Again, Ellsbury's bat managed to absorb virtually all the ball's energy, nudging a 90 mph pitch just a couple feet away. Yet the crowd didn't groan with disappointment, but roared in anticipation, naturally comparing this with the previous day's exemplar. But this particular hit just wasn't crappy enough, and for those of you scoring at home, it went down as your classic ground out to catcher, to retire the side.

Oh right. Just 1 game out.


Micah said...

interesting use of the word "archetype." in the study of the mental representation of categories, (see Murphy's 2002 "the big book of concepts" for an extensive review, in case some one really cares) you contrast "archetype" with being the most typical, or basically the central tendency of the category, what has been called the "prototype." the common example of this, a robin is pretty much a statistical average of all birds, and so is considered most typical. or for car, like a toyota camry. and then how "good an example" of a category something was, had to do with distance from this central tendency. so, a camry is a better example of a car than a ferrari, even if a ferrari is a better car. however, people have also researched illustrative extreme exemplars, and their importance. the first time this became relevant was for barsalou's (1983,1985) work on goal-derived categories. and here, goodness of example has more to do with the goodness of achieving the goal, so the more extreme the better, not the more average. a classic example is "diet food" where the best example would be a food with 0 calories, and then the goodness of example would be based on distance from 0. the average diet food, and actually every diet food, has more calories. barsalou calls these "ideals"
another related notion is what goldstone (1996) called "caricatures." unlike "ideals," these "best example" of categories aren't the most extreme, but they are pushed in that direction from the central tendency in order to be distinguished from related categories. if categories have overlapping distributions, and are learned together, then a "psychological" separation, is useful, perhaps in generating explanations. for example, to explain how an "ellsbury hit" is different from another hit, a more extreme example will be more illustrative.
now, one way prototypes have been shown to be important is for supporting property induction to other category members (osherson et al 1990). so, let's say someone heard about a new bird, knew very little about it, and wanted to determine if it had property X. then let's say one was told that either robins had property x, or they were told penguins had property x. in the first case, one would be more confident that this new bird had property x than if they were told the second. there hasn't been the same work on induction from ideals and caricatures (the constructs related to your "archetype"), but for now we'll assume they play the same role in induction as prototypes do in their respective categories.

Now, we can get to the crowd's faulty induction, but unfortunately, more technical category jargon will still be needed. the crowd reaction to the ellsbury squibler is that they were reminded of this archetype, and generalized. the category "ellsbury hit" is a caricature or ideal because it emphasizes these extreme values. here an example was even more extreme along this dimension, it traveled even slower, and went an even shorter distance, making it even more identifiable as an ellsbury hit compared to a hit by anyone else, so seemed likely to be one. but here, there was a faulty assumption of linear separability between the categories "ellisbury hit" and "ellsbury out". if one were to create a space, and divide the space into two categories, one could do so with a single line, or not. if one does so, the categories are "linearly separable." for example, if there was a space defined by the dimensions velocity and displacement from home plate for all possible balls hit into play, then one could potentially draw a line through this space and say "on this side of the line, it is a hit, and on this side of the line, it is an out." so, for ellsbury squiblers, perhaps the crowd hypothesized, all balls hit into play with less speed than value X, and move as little from home plate, or less, as value Y, then it is a hit. however, this is not the case. a hit is a category defined by a sweet spot in the space, not a line though it. meaning, there is a limited range of values in each direction that will still be a category member. it's can't be too fast, or too slow, it has to be just right.
there has not been work on induction in "sweet spot" categories, or on the folk assumption of linear separability in natural categories, (i think), but maybe i'll do it.

Soxlosophy said...


thanks, micah!