back |
Dan Lloyd
The MIT Press, 2004
There’s a standard in detective shows—bring in the Profiler, get inside the criminal’s head, root out that psycho-consciousness. “See those bite marks, that misplaced shoe-tree. We’re looking for someone who loves their mother.” So when Dan Lloyd—academic philosopher—wants to write “a novel theory of consciousness” he turns to crime fiction, the natural genre of phenomenology, and writes Radiant Cool, a novel that starts with a murder and ends with a reference list—a couple of hundred pages of crime fiction, then a hundred or so of explanatory notes.
The first two-thirds of Radiant Cool are a crime-noir frame for Lloyd’s take on consciousness. Page one and the disgruntled, always wears black, Miranda Sharpe finds the body of Max Grue, her PhD supervisor. The body soon disappears, leaving Miranda to search for Grue in a series of set pieces that are sometimes a bit clunky stylewise, but mainly fun and pacy. As with the noirs of old there are lots of deviously motivated strangers ready to make friends with, and later betray, the heroine. Guns go off, computers appear to crash, world domination is thwarted. And in classic noir style it doesn’t end well for our heroine either.
Overt, oblique and insider references abound. This is not a novel of character development; the characters are there for symbolic reasons, complete with loaded names. The hardcore psychiatrist, Clare Lucid (the bravery of cornball jokes), is a parody of mind-as-software types—who needs wetware, the mind could just as easily run on a suitably organised collection of sandwiches. When Miranda fakes a problem for an excuse to see Lucid the first few lines of the therapy session mirror the output of a session with ELIZA, an AI program written in the 60s that faked being a Rogerian psychotherapist.
MIRANDA: I don’t quite know where to begin.
LUCID: Is it because you don’t quite know where to begin that you...
The Russian detective is Porfiry Petrovich Marlov (the detective in Crime and Punishment plus Philip Marlowe). The scientist, Zamm, sees the brain like an engineer’s block diagram (he’s named after an applied maths journal). The name of the missing Grue comes from a famous problem in induction set by Nelson Goodman—grue refers to a property that depends on time. Max Grue’s big insight is that consciousness depends on time. Grue is also a particularly self-absorbed philosopher and his death, blinded by stroke, unable to contact others, rambling on deep within his own subjectivity, is strictly Liebniz’s “windowless monads.”
On top of the insider jokes (there’s a great reference to Young Frankenstein, plus suitably nerdy names for computer systems) are great chunks of explanation that act as background to the processes Lloyd has gone through to develop his theories. Marlov teaches Miranda about multidimensional scaling, a method for visualising complex data sets. A lot of space is devoted to how Jeff Elman’s recurrent neural nets encode time and context. It all sits fine within the novel and would sit just as well in an introductory textbook.
The various explanations get elaborated in the final non-fictional section where Lloyd gets more formal, puts phenomenology first, neuroscience second, and uses stats as the great decider. Makes for a nice change from the great mass of books that use—shudder—quantum physics to explain where all the juju mind stuff goes on. Lloyd works through defining consciousness then tackles some explanation to material cause. Start with superposition—the way that perceptions can have lots of interpretations all at once. Is it a bird, is it a plane...how can one sensory input be so loaded with possibility? From neuro the explanation is that superposition follows from the way experience is coded in the brain—sensory inputs get recorded strongest on the path they come in on, gradually mixing it up with traces the further away from the input pathway they get. Elaboration and contextualisation. Traces are bidirectionally connected so that activating one activates others according to the strength of their past association. Seeing “is it a bird” activates neural traces corresponding to all previous “is it a bird” experiences, and traces are being activated by the other stuff that is going on not directly connected to the visual input. Activation spreads, priming indirectly related traces—say the activations that occur when seeing a plane overhead. Those activations are ready to go but not up to threshold yet. With the sound of a plane new neural patterns are strengthened, others weakened, and “it’s a plane” pops out. Superposition as a traversal through plausible cause.
But the really big deal for Lloyd is Husserl’s idea of temporality, how consciousness has to be able to link events together so that we hear a melody rather than a succession of discrete and unrelated frequencies. Lloyd reasons that if consciousness is in the brain then temporality should show up in the unfolding of neural activity in time. Activity will be similar to itself in the short term, but always moving forward, never to repeat. He finds evidence in functional magnetic resonance images taken of people performing a range of experimental tasks. He runs the stats, dynamical systems and multivariate scaling as per the novel, and finds what he wants, never the same brain twice but always most similar to itself when closest in time. Lloyd takes this as evidence for temporality and, by extension, consciousness. However he has only shown that brains slowly change and maybe slowly changing is just the way brains are, conscious or not. Learning and experience must change the brain but they can happen without consciousness. Getting a scan, head stuck in a magnetic field, hydrogen atoms synching up and spinning out of phase as the blood’s perfusion of the brain slowly changes—that fits the ‘learning experience’ tag.
Does Lloyd have a novel theory and compelling argument? Not really, but with Radiant Cool he does give a nice intro to the field of consciousness studies. After which try Walter Freeman, Francisco Varela, Tim van Gelder and others.
RealTime issue #68 Aug-Sept 2005 pg. 34
© Greg Hooper; for permission to reproduce apply to [email protected]
