In a fascinating paper, Prof Haselager provided a forensic analysis of what it means to ‘read’ and examined whether neuroscience really does ‘read’ the brain. Reading the mind, he proposed, might be akin to reading messages (like reading Facebook). The messages read take the form of desires, intentions, thoughts, hunches, dispositions, qualia. Mind-reading certainly is not the unitary activity that brain-reading appears to be.
Haselager noted that neuroscientific techniques might work for some of these diverse ‘states’ and not for others. Where neuroscience does shed light on the brain, Haselager was interested to know, what kind of light this is. For example, does brain-reading decode information that is coded, an activity that seems to be a central function of reading? Haselager proposed that in essence brain-reading is a two stage process. First, neuroscience looks at brain-activity and correlates that to cognitive processes. Second, it infers from brain-activity, so understood, that a cognitive activity is taking place.
He examined this two-stage process through the important notion of representation. Two central elements of representation are the ‘vehicle’ and the ‘content.’ The vehicle is akin to a physical carrier of content. Ink might be such a carrier, or paper, or the airwaves produced when I open my mouth to talk. Representation consists in the vehicle and its content.
Haselager asked of FMRI techniques whether they ‘understand’ content-meaning, and suggested that content-wise such techniques are opaque. He used the analogy of ‘reading’ a pin-number. What is the pin-machine doing with the code? Content, for Haselager, has nothing to do with my communication with the pin-machine. It receives an opaque instruction that allows it to identify the correct vehicle but not to understand the content. This is precisely the limitation of brain-reading as Haselager sees it. The current stage of neuroscience does not enable it to read content. Rather it allows us to categorise patterns of vehicle similarity. Even Haynes, he notes, observes that it is not possible directly to read the language of the brain. In essence brain-reading does not allow us to read thoughts; it gives no systematic access to propositional attitudes. Vehicle-mapping is not really brain reading. It is not decoding mental content. It follows that if a legal process demands access to content, it ought to be cautious in how it uses neuroscience.
The paper was very much thought-provoking in considering what precisely it is that neuroscience does not quite do.
Comments