In the most recent "Room for Debate" feature in The New York Times, five scholars involved in law, psychiatry, and criminal justice discuss the current state of the insanity defense in light of the Tucson shootings. A nice range of viewpoints is presented, from Alan Dershowitz's argument that the insanity defense has been marginalized, politicized, and abused, to James Q. Whitman's lament that, compared to the Europeans, the American legal system is much less sympathetic towards claims of insanity due to a stronger emphasis of individual responsibility and (therefore) retribution, to David I. Bruck's plea for keeping the insanity defense based on individual responsibility and culpability.
Perhaps my favorite two contributions to this debate were from William T. Carpenter, Jr. and Kent Scheidegger, who both question the role of medical and psychological science in informing juries decisions regarding insanity and its relationship to guilt. This brought to mind my favorite piece of scholarship on the insanity defense, on which I focus in my senior seminar on crime and punishment: Christopher Slobogin's "An End to Insanity: Recasting the Role of Mental Disability in Criminal Cases" (Virginia Law Review, 86/6, Sept. 2000, 1199-1247). Slobogin, one of the leading experts on mental disability and the law (see his books here), argues that the insanity defense as a specific defense should be eliminated, and instead evidence of mental defect or impairment should be used to support general defenses such as mistaken belief or absence of mens rea. Since it is not insanity per se, but only how a defendant's mental state influenced his or her actions, that impacts on responsibility, it more properly enters the determination of responsibility through one of the other standard defenses, not as an independent defense detached from more traditional ones (and therefore inviting more controversy).
For example, lets's say Bob mistakenly believed, because of mental illness, that the police officer he shot was his romantic rival trying to kill him. Here, Bob's mental illness is relevant to his act because it influenced the mistaken belief that led to the act, not simply because he suffers from mental illness. It is in this sense that Slobogin argues that the insanity defense is "overbroad" (1202); insanity is not always relevant to a defendant's actions, so it should be invoked only when it contributes to one of the standard excuses that bar culpability, not as a sort of blanket excuse which invites the type of abuse Dershowitz discusses.
Furthermore, along the lines of Bruck's comment, Slobogin's argument for eliminating the insanity defense (but not the impact of mental illness in criminal trials) is based on culpability and retributivism, because he wants to ensure that people are convicted of a crime if and only if they are truly responsible for it. And rather than weakening consideration of mental illness in criminal cases, his proposal would likely make it more likely to be considered because the "spectre" of the insanity defense and its purported unfairness would no longer loom over the criminal justice system and obscure its true purpose.