In lieu of an abstract, here is a brief excerpt of the content:

The Greek god Momus is said to have expressed dissatisfaction with human beings because our state of mind cannot be readily discerned. It would have been better, Momus thought, if we were born with a window into our breast so that our mental states could be easily recognized. Developments in functional neuroimaging may be starting to provide the kind of window that Momus desired, for in certain contexts functional magnetic resonance imaging data can provide useful clues to a person’s state of mind. This chapter considers the implications of such developments for questions concerning the nature of human freedom and autonomy. Let us begin by considering some of the limitations of decoding (or socalled “brain-reading”) research (Bayne, 2012). There are three points to note. Firstly, to date decoding has been carried out only in carefully controlled laboratory contexts in which the range of intentions that individuals are likely to have is artificially restricted. Consider, for example, the study in which Haynes and colleagues (Haynes et al., 2007) were able to predict with up to 70% accuracy whether subjects would add or subtract two numbers that had been presented to them. Because the subjects had been specifically instructed to either sum the numbers or to subtract the numbers, the experiments had only to decide which of two possible intentions subjects had decided to adopt. It is very much an open question whether the level of “decoding” accuracy that they were able to obtain in this highly constrained environment can also be obtained in everyday contexts in which the range of intentions that subjects can form is essentially unbounded. A second point to note is that decoding will typically be useful to us only when it can be reliably carried out at a fine-grained level of analysis. Decoding techniques that enable us to determine that a person is (say) thinking about blowing up planes rather than cooking asparagus are unlikely to be of much interest if they cannot also tell us whether a person 4.1 Neural Decoding and Human Freedom Tim Bayne 178 Tim Bayne intends to blow up a plane, desires to blow up a plane, fears that a plane will be blown up, or is merely entertaining the possibility that a plane might be blown up. Nor will they be of much use to us if they can’t distinguish the intention to blow up a plane from the intention to prevent a plane from being blown up. This is an important point, for it is likely that the reliable decoding of fine-grained mental states will generally be much more difficult than the reliable detection of coarse-grained mental states. Thirdly, decoding techniques do not really provide us with the window into the soul for which Momus hankered, for in order to use functional neuroimaging information to ascribe mental states to an individual we need to know the correlations between functional neuroimaging states and mental states, and identifying such correlations requires that we treat verbal reports and other behavioral measures as accurate markers of mental states. In other words, the sense that decoding techniques provide us with direct access to an agent’s mental states is an illusion, for the practice of using neuroimaging data to ascribe mental states is as reliant on inference and interpretation as the practice of using behavior to ascribe mental states is, if not more so. With these cautionary remarks in place, let me turn now to the ways in which decoding research might bear on questions concerning human freedom and autonomy. There is no doubt that decoding research—and indeed cognitive scientific investigations into human agency more generally —is widely taken to threaten our self-conception as free agents (see, e.g., Hallett, 2007; Halligan & Oakley, 2000) But are such threats wellmotivated ? How exactly might “brain reading” threaten to undermine the idea that we are free and autonomous agents? One perceived threat centers on what we might call the “libertarian assumption”—the assumption that in order to be freely willed an action cannot have a fully sufficient cause. Cognitive science might be thought to threaten free will insofar as it suggests that neural states provide fully sufficient causes for all human action. Although it is doubtful whether cognitive science has shown—or even could show—that all human actions have fully sufficient neural causes, I will waive that concern here. Instead, the more interesting issue is whether the libertarian assumption is justified. There is little consensus on this point...

Share