Friday, August 18, 2006

On the Detection of Agency and Intentionality in Nature

AUTHOR: Elena Broaddus

SOURCE: Evolution and Design

COMMENTARY: Allen MacNeill

First, many thanks to the faithful readers who have also continued to pay attention to the Evolution and Design website (the weblog of the "notorious Cornell evolution and design seminar" and the contents contained therein. I am particularly pleased that the hard work and careful thought of the students whose papers have been posted has been recognized, and even moreso that they have been given the highest praise possible: that is, critical analysis.

I would like to draw some more attention to E. Broaddus paper on the “innate” tendency to infer purpose in nature. I have long suspected that humans (and perhaps many vertebrates, especially mammals) have this tendency. As an evolutionary psychologist, I at least partially subscribe to the idea that the human mind is composed primarily of “modules” whose functions are to process particular kinds of sensory information in such a way as to yield adaptive responses to complex environmental information. This is precisely what Broaddus argues for in her paper: that the human mind (and, by extension, the vertebrate “mind” in general) has a module that is adapted specifically for the precise and rapid inference of intentionality in nature. That such an “agency detector” (to use the commonly accepted term for such a module) would have immense adaptive value is obvious. In an environment in which other entities do indeed have “intentions” (i.e. predators, competitors, potential mates, etc.), the ability to detect and infer the possible consequences of acting upon such intentions would confer immense adaptive value on any organism with such an ability.

Furthermore, as Broaddus points out (and as we discussed briefly in the seminar), to be most effective such a detector should be tuned in such a way as to detect virtually all such “intention-indicating” behaviors. This would have the effect of producing a significant number of “false positives,” as any detector that is tuned high enough to detect all actual cases would have such a side-effect.

As Broaddus points out, one of the side-effects of such an “agency detector” would be the detection of intentionality in entities that clearly had no such intentions. If, for example, one of the most important functions of such a detector in humans is to quickly “read” and assess the intentions betrayed in human facial expressions, then it would almost certainly detect human facial expressions in objects in the environment that clearly do not have such expressions, such as rocks, foliage, water stains, etc. This would explain the ability of many humans to “see” human facial expressions in such things as water stains, cinnamon buns, rocks, etc.

Clearly, there are some “natural objects” that do, indeed, have human facial expressions impressed upon them: the faces of the presidents at Mount Rushmore are an example cited ad nauseam by ID theorists. However, I am much more interested in “faces” that humans detect in rocks and other environmental objects that are clearly not produced by human agency. Indeed, the faces at Mount Rushmore constitute a kind of “control” for this ability, as they are clearly the result of intentionality, and therefore can be used to anchor that end of the “agency detection” spectrum (at the other end of which are things like “faces” in clouds, tree foliage, etc.). Somewhere in this spectrum is a cross-over point at which actual intentionality/agency disappears and facticious intentionality/agency takes over. It is the location of that cross-over point that constitutes the hinge of the argument between evolutionary biologists and ID theorists.

Broaddus’s analysis of autism as a possible example of malfunctioning “agency detection” is, IMO, brilliant, and presents an immediately testable hypothesis: that autistic children lack well-tuned “agency detectors,” and that this at least partially explains their well-known indifference to intentional agents, such as other people (including their parents), animals, etc. In people with both full-blown autism and the milder Asperger’s syndrome (sometimes called Aspies”), a common attribute is an impaired ability to infer intentionality (or, in many cases, the mere existence of other minds) on the part of autistics and Aspies. As Broaddus points out, there are clear anatomical and functional differences between autistics, Aspies, and non-impaired people, and that these differences may be correlated with the etiology of these conditions. For example, it is very interesting that there appears to be more (rather than less) neurons in the brains of autistics than in non-impaired people.

This lends credence to the generally accepted hypothesis that the information processing “modules” proposed by evolutionary psychologists are the result of “pared down” neural networks that are speciallized for particular cognitive tasks. Clearly, the agency/intentionality detector in humans functions extremely well and, as the parlance goes, “in the background.” We are rarely conscious of its operation, despite the fact that it is virtually always “on.” This explains, for example, something I first noticed as a young child: that no matter how much I tried, I couldn’t NOT see faces in the patterns in the linoleum on the floor of my grandmother’s kitchen, in the foliage of trees, in rocks, and in photographs of billowing smoke, splashing water, etc. The agency/intentionality detector works extremely efficiently in people of all ages, but especially in children. Indeed, as Broaddus points out, part of becoming an adult consists in learning (usually by trial-and-error) which of the seemingly intentional entities which we perceive all the time actually are intentional agents and actually have intentions vis-a vis ourselves. We must learn, in other words, to critically analyze the constant stream of “positive” agency/intentionality detection events, and discriminate between those that affect us and those that do not. It may be that this discrimination process actually involves the neurological “re-wiring” of the parts of the sensory/nervous system that produces such detection events, and this might explain, at least in part, the decreased ability of adults to believe in the existence of intentional agents in the natural environment.

Broaddus not only presents a cogent hypothesis concerning the existence of such an agency/intentionality detector/module in humans, she proposes several possible ways of testing whether or not such a detector actually exists, and to “map” its dimensions, capabilities, biases, and limitations. I believe that this opens up a very fruitful area of empirical research into such detectors, and can ultimately lead to much more clarity about an issue that so far has generated much more heat than light. I hope that her ideas and suggestions will be followed up by others (I certainly intend to do so), and that further empirical research into this fascinating and little-known capability will add to our understanding of what makes us the peculiar creatures we are.

Labels: , , , , , ,


At 9/03/2006 02:48:00 PM, Blogger teleologist said...

Prof. Allen MacNeill has been arguing on his blog, that Intelligent Design Theorists (IDT) have fallen into the fallacy of inferring design to objects that in actuality, have no teleology at all. He hypothesizes the vertebrate mind is composed of modules. One of these modules is a highly effective intentionality detector.

In support of this argument, he uses an example from his student’s paper in the Cornell ID class. Allen writes,

As Broaddus points out, one of the side-effects of such an “agency detector” would be the detection of intentionality in entities that clearly had no such intentions. If, for example, one of the most important functions of such a detector in humans is to quickly “read” and assess the intentions betrayed in human facial expressions, then it would almost certainly detect human facial expressions in objects in the environment that clearly do not have such expressions, such as rocks, foliage, water stains, etc

Allen continues his argument with this example. (more…)

At 10/15/2006 02:28:00 AM, Anonymous Anonymous said...

Perhaps you have already received this feedback - but Broaddus is reporting very old news. Simon Baron-Cohen was writing about the link between autism and a defective "intentionality detector" module in 1995 (see "Mindblindness," 1995 MIT Press). A huge literature as since arisen around this topic, as well as the more general notion that the detection of intentionality is an evolutionary adaptation.

At 10/15/2006 11:43:00 AM, Blogger Allen MacNeill said...

Hi, Bill: Yes, I am aware of Baron-Cohen's work on "mindblindness" among autistics. However, it was my impression that this referred primarily to the apparent inability of autistics to recognize or respond to the presence of "other minds" in other people.

This is related to, but significantly different from the inability that Broaddus proposes in her paper: that autistics have a defective "agency detector." True, minds are agents, but she (and I) are more interested in the detection of agency in nature, not other people. If autistics, for whatever reason, cannot use the mental machinery that most of us use to detect the presence of other minds in other people, perhaps they also cannot detect agency in general, including agency acting in nature.

An interesting question, and one that I don't believe that Baron-Cohen addressed this aspect of the problem at all, but I intend to check his research (and subsequent work by other researchers) to see if I'm mistaken. Thanks for the tip!

At 10/15/2006 01:36:00 PM, Anonymous Anonymous said...

Hello again, Reciprocating Bill here -

I agree that the Baron-Cohen's detector is a subset of the detection of agency. Agency entails the ascription of mind-like qualities he describes, as well as the detection of attention, intention, and, at higher propositional levels, beliefs, desires,etc. As you are no doubt aware as an evolutionary psychologist, there are huge literatures both in primatology and developmental psychology regarding the evolutionary and ontogenetic emergence of these linked capabilities - probably somewhat misnamed "theory of mind" when "theory of agency" may be more appropriate. I think of Whiten and Bryne's influential Machiavellian Intelligence thesis, Michael Tomasello's work, Brian Hare's interesting experiments with various primates (and dogs), Daniel Povinelli, Wellman, Meltzoff, Andrew Whiten, and so on. Huge, and hot. (This was the general neighborhood of my 2004 dissertation.) And I agree that this is likely the basis of the projection of agency onto nature and the deployment of spiritual notions (essentially agency writ too large). In fact, the working title of my dissertation, and now my extremely neglected blog, is "The Structure of Angels," hinting at this last point.

BTW, I am checking my watch and timing your ejection from Uncommon Dissent. Won't be long, as you are pressing your points much too effectively.


Post a Comment

<< Home