… one must make a distinction between ontological reduction and explanatory reduction. This distinction is a commonplace in the philosophy of science, but it is routinely ignored in the hype surrounding cognitive neuroscience.
That quote is from an article, The Limits of Neuro-Talk by Matthew B. Crawford in the winter 2008 issue of The New Atlantis. He goes on to say:
The error goes like this: from the fact that some phenomenon is composed of and dependent upon more fundamental parts, it is thought to follow that any explanation of the higher-level phenomenon can be replaced by, or translated without residue into, an explanation at the lower level of its parts. Once this reduction is (putatively) accomplished, the ontological status of the higher-level phenomenon is demoted to that of mere phenomenon: appearance versus reality. Our gaze is shifted away from the thing we initially wanted to understand, to some underlying substrate. This procedure is thought to be enjoined by the conviction we all share with the natural scientist: there is only one universe, and it is made up of physical particles.
Yet the natural scientist knows just as surely that our best account of that universe is, in many cases, not forthcoming from physics. We turn instead to chemistry or biology. The need for such “special” sciences that take higher-level structures as given does not compromise the bedrock ontological supposition that there is a single universe, made up of physical particles. One can have one’s materialism while admitting the autonomy of higher-level disciplines. There is much confusion on this point, and it seems to be bolstered by a fear that to be less than completely reductive in one’s explanatory posture somehow commits one to “spiritualism.”
The explanatory independence of biology, its irreducibility to physics, is consistent with biological entities being composed of and dependent upon physical entities. The biologist believes that the dog is made up of nothing but protons, neutrons, and electrons, but he does not try to give an account of the dog at that level. Is this merely due to the limitations of our current state of knowledge? Would it be possible in principle to construct a comprehensive understanding of the dog starting from particle physics? The consensus view appears to be that it is not possible even in principle, due to considerations of complexity and non-linearity, or thermodynamic irreversibility (take your pick). Even within physics, lower-level accounts sometimes presuppose structure that is identifiable only at a higher level, or depend upon boundary-conditions that cannot be generated from within the lower-level account. Even something as simple as a volume of gas displays “emergent properties” (here, an irreversible tendency toward equilibrium) that cannot be derived from the collisions between individual gas molecules (which are symmetric with respect to time).
Earlier in his essay, by way of explanation, Crawford writes:
… attention is thought to be a function realized in some brain region. But this correction does not vacate the force of Uttal’s criticism, because functions, like properties, are distributed (they require a whole system or mechanism to be realized), and one must pause to ask: what are the boundaries of the pertinent system? A danger inherent in the localization thesis may be illuminated by analogy to an internal combustion engine. In describing an engine, one might be tempted to say, “the opening of the intake valve is caused by the movement of the rocker arm.” Except that the rocker is, in turn, set in motion by the camshaft, the camshaft by the crankshaft, the crank by a connecting rod, the rod by the piston. But of course, the piston won’t move unless the intake valve opens to let the air-fuel mixture in. This logic is finally circular because, really, it is the entire mechanism that “causes” the opening of the intake valve; any less holistic view truncates the causal picture and issues in statements that are, at best, partially true. Given that the human brain is more complexly interconnected than a motor by untold orders of magnitude, it is a dubious undertaking to say that any localized organic structure is the sufficient cause and exclusive locus of something like “reason” or “emotion.”
… Those who would use science to solve real human problems often must first translate those human problems into narrowly technical problems, framed in terms of some theoretically tractable model and a corresponding method. Such tractability offers a collateral benefit: the intellectual pleasure that comes with constructing and tinkering with the model. But there is then an almost irresistible temptation to, as E. A. Burtt said, turn one’s method into a metaphysics—that is, to suppose the world such that one’s method is appropriate to it. When this procedure is applied to human beings, the inevitable result is that the human is defined downward. Thus, for example, thinking becomes “information processing.” We are confronted with the striking reversal wherein cognitive science looks to the computer to understand what human thinking is.
In a companion piece Neuroimaging and Capital Punishment by Carter Snead also in the winter 2008 issue of The New Atlantis, we find a conundrum:
… In the context of sentencing, desert and dangerousness inevitably conflict. “To advance one, the system must sacrifice the other,” as Paul Robinson has observed. “The irreconcilable differences reflect the fact that prevention and desert seek to achieve different goals. Incapacitation concerns itself with the future—avoiding future crimes. Desert concerns itself with the past—allocating punishment for past offenses.” This tension is played out in dramatic fashion in capital cases. On the one hand, capital defendants introduce mitigating evidence to diminish their moral culpability, thus seeking a final refuge in the concept of retribution. On the other, the prosecution tenders evidence of future dangerousness, trying to stoke the consequentialist fears of the jury about violent acts that the defendant might commit if he is not permanently incapacitated by execution. In capital sentencing, pure consequentialism is the gravest threat to the defendant’s life, while appeals to retributive justice are often his last best hope.
… In fact, the widely shared intuition that seems to be motivating the long-term aspiration—namely, that retributive justice is the primary source of the brutality and harshness of the modern American criminal justice system—may generally be misguided. As Paul Robinson has argued, “the harshness of the current system may be attributed in largest part to the move to rehabilitation, incapacitation, and deterrence, which disconnected criminal punishment from the constraint of just desert.” Many features of the criminal justice system that are frequently criticized as draconian and inhumane are, in fact, motivated by a purely consequentialist crime-control rationale. Such measures include laws that authorize life sentences for recidivists (such as the “three strikes” laws), laws that reduce the age at which offenders can be tried as adults, laws that punish gang membership, laws that require the registration of sex offenders, laws that dramatically increase sentences by virtue of past history, and, most paradigmatically, laws that provide for the involuntary civil commitment of sexual offenders who show difficulty controlling their behavior. These laws are the progeny of the principle animating the long-term aspiration of those wielding neuroscience in capital cases, and some are worrisome examples of its possible implications.
Robinson points to the possibility that “if incapacitation of the dangerous were the only distributive principle, there would be little reason to wait until an offense were committed to impose criminal liability and sanctions; it would be more effective to screen the general population and ‘convict’ those found dangerous and in need of incapacitation.” Indeed, the short-term project—using cognitive neuroscience to identify the roots of criminal violence—may someday create novel and powerful opportunities to interfere with individual liberty.
…In the end, while the goals of making capital sentencing more rational and humane are laudable, the cognitive neuroscience project to do so is ill-conceived. Because that project’s short- and long-term aims are intractably opposed to one another, its impact likely will be the opposite of what it seeks. It is unclear whether it is possible (or desirable) to salvage the project in a way that will preserve its humanitarian ends. It may be that the reductive materialist account of human personhood and human agency posited by cognitive neuroscience—and, indeed, by modern science more generally—is fundamentally incompatible with the account on which the criminal law is premised. This should lead us to question the assumptions underlying cognitive neuroscience no less than those underlying the criminal law. If we examine both through the lens of our humanitarian aspirations, we are likely to discover that the wisdom behind our laws fares a good deal better than we imagined against the assumptions (often masquerading as hard facts) behind the new science of the brain. Not surprisingly, dehumanization turns out not to be humane.
There’s much to think about in both these articles. I’m (obviously) not knowledgeable enough to judge the merits of what they are saying, but they seem to make good arguments for their positions.
Both articles are very long, but highly recommended, if you have the time.
The Limits of Neuro-Talk
Neuroimaging and Capital Punishment
-Julie
