Saturday, March 14, 2009

The Near-Term Prospects for fMRI Lie Detection

2 Margaret A. Boden, MIND AS MACHINE 1227-1228 (Oxford: Clarendon Press 2006):

Linking rats or monkeys to robots is a major exercise, not undertaken lightly. But brain-scanning studies on humans are less tricky, and are multiplying merrily. ... As for the professional journals, by the time you read this book they will have carried thousands of PET/fMRI reports.

Their theoretical significance, however, is debatable. Brain imaging has even been dubbed "a neo-phrenological fad", because of the difficulty of interpreting it in terms of psychological functions (Uttal 2001). There are four main problems.

[snip, snip]

And fourth, one can't sensibly suggest just what's being done by the high activity (even if one knew it was excitatory activity) without a theory at the cognitive/psychological level, specifying just what computations might be involved when the thought in question occurs. Usually, no such theory is available. ... In short, most brain imaging is an a-theoretical fishing expedition: more natural history than science. As Gazzinga had put it, before this new 'industry' burgeoned, neuroscience needs cognitive science.

Responsible researchers know all this, of course, and are careful. But the irresponsible ones--and a fortiori the journalists--seemingly don't, and aren't.


the dynamic evidence page

coming soon: the law of evidence on Spindle Law

Friday, March 13, 2009

The Sophistication of (Some) Common Sense

In thinking about juries, jurors, lay knowledge, and lay participation in the legal process, it is worth thinking about the implications of insights such as the following:
Men are in a more difficult intellectual position than Life robots. We don't know the fundamental physics of our world, and we can't even be sure that its fundamental physics is describable in finite terms. Even if we knew the physical laws, they seem to preclude precise knowledge of an initial state and precise calculation of its future both for quantum mechanical reasons and because the continuous functions needed to represent fields seem to involve an infinite amount of information.

This example suggests that much of human mental structure is not an accident of evolution or even of the physics of our world, but is required for successful problem solving behavior and must be designed into or evolved by any system that exhibits such behavior.

John McCarthy, "Ascribing Mental Qualities to Machines" (1979)" (with updates by author here & there)

the dynamic evidence page

coming soon: the law of evidence on Spindle Law

Tuesday, March 10, 2009

The Riddle of Non-Individualized Forensic Expertise

Well, that -- the topic in the header -- is a real mouthful. It is nevertheless the problem I want to say a few words about.

In my last post I quoted the Texas Court of Criminal Appeals. That court said:

Judge Womack contends that our exclusion of expert conclusions concerning the truthfulness of allegations is illogical. He isolates three different types of statements that an expert might make: (1) children who are fantasizing or being manipulated behave in certain ways, (2) this child did not behave in those ways, and (3) this child was not fantasizing or being manipulated. He claims that (3) necessarily follows from (1) and (2), and, because we permit expert statements of the type (1) and (2) variety, we must necessarily permit conclusions of type (3). But, for his argument to work, Judge Womack must assume that the expert testifies that all children who are fantasizing or being manipulated behave in certain ways. If only some exhibit the behaviors in question then a conclusion that a particular child is not fantasizing or being manipulated because he does not exhibit the behaviors does not necessarily follow. But, given the imprecision of psychological science and the variability of human nature, no competent and honest expert could make the global statement necessary to satisfy Judge Womack's syllogism. At most, statements of type (1) and (2) would provide some inductive support for a conclusion of type (3). But an expert opinion of type (3) would also be supported by personal observations and lay knowledge of human behavior. The latter is clearly the exclusive province of the jury, and the former, along with statements of type (1) and (2) can be imparted to the jury by the expert. The jury can then make the inferences necessary to determine whether a type (3) conclusion is warranted without the expert commenting on that issue.
Here's the riddle: If the expert's expertise is relevant to the issue at hand (e.g., "Was this particular eyewitness identification accurate or inacurate", "Does this particular witness suffer from delusions?", "Does the syndrome evidence show that this particular child probably delayed reporting because of embarrassment rather than for another reason?", and so on), why isn't the expert not only permitted but required to give an opinion about the behavior of the specific individual? Is it because the expert has no expertise about the specific individual? Well, no that cannot be -- for then the expert's expertise would not be relevant to the actual issue at hand. Is it because the expert would invade the province of the jury by giving an opinion? But courts have already crossed that imaginary Rubicon by allowing the expert to testify, haven't they? Well, then, is it because the expert doesn't know all the evidence and facts about the individual that the jury knows? Well, that's not going to explain things, is it, if the expert has sat through the entire trial and knows everything (and more) than the jury knows. So what is the explanation? We can't very well say (as the Texas court seems to suggest) that it's because the jury knows better than the expert how to combine different kinds of information (or, as some would put it, different reference classes). If that were the explanation, it would require the premise that the experimental evidence of the expert failed to take into account relevant variables or, stated differently, that the expert's expert knowledge does not speak to the individual, particularized issue at hand.

Isn't it the case that the position that courts take -- let the expert testify in general terms (about, e.g., factors that affect the accuracy of eyewitness identification in general) but not in particular terms (e.g., about the accuracy of this witness' identification) -- is an unprincipled compromise and muddle, one that straddles the fear of doing without helpful information and the fear of allowing the decision in a case depend on unhelpful or irrelevant information?

the dynamic evidence page

coming soon: the law of evidence on Spindle Law