Month: May 2008

This weekend I read three very different books on the human mind and brain.
Jeff Hawkins On Intelligence poses a speculative theory about how the neocortex works. Hawkins asserts that the distinctive aspect of human intelligence is that it allows us to make predictions. Based on a few strands of previous research, some insight, and aggressive reverse engineering, Hawkins proposes a neural architecture that enables humans to generalize patterns from raw sensation, allowing us to predict the next notes of a familiar song and to extend knowledge with analogies.
The hypothesis about how the neocortex works is interesting. The way that it proposes the generation of predictions by a combination of top-down and bottom up feedback is clever. The observation that sense-making requires a dimension of time — not just music, but touch and sight — is insightful. Unlike the evolutionary hypothesis of say, Terrence Deacon on the origins of symbolic thinking, Hawken’s algorithm is testable. However, Hawkins’ understanding of intelligence leaves out some crucial factors. Hawkins is interested in the mind as disconnected from emotions and desires. He believes that computers that have predictive intelligence without ambition, lust or greed will have the good of human intelligence, without the flaws introduced by the passions.
This dualistic vision ignores the insights of Antonio Damasio, a neurologist whose theory of intelligence embraces the emotions. Damasio observes patients with injuries to emotional processing, and finds that they lack the senses of fear and anticipation that enable people to make functional decisions. A lack of normal empathy prevents someone from getting along with other people. A computer that implemented predictive learning without emotions might be some combination of sociopathic and unwise. A computer that implemented learning without boredom and forgetting might not even be optimally effective at synthesis.
Hawkins focuses on the connections between neocortex and senses, but ignores the connections between neocortex and emotional parts of the brain. The neuroscience bits of Deacon’s book explain how in humans, the connections between the limbic system and the neocortex became intertwined as humans evolved. There’s a biological basis to Damasio’s observation that emotions are part of intelligence.
Where Hawkins focuses on human intelligence as a superb prediction engine, Gary Marcus focuses on the flaws and glitches in human smarts in areas such as decision-making, language, pleasure-seeking, and mental illness. In Kluge, the Haphazard Construction of the Human Mind, Marcus counters against evolutionary and anti-evolutionary arguments that the human mind reflects the best of possible worlds. Instead, the mind is a hodgepodge awkwardly cobbled together.
We don’t do a good job of making decisions about financial risk, or resisting temptation, because of our biological tendency to maximize short-term gain. Here, Markus shares a bias with Hawkins, that reason would result in better outcomes. But if you eliminated the motivations of hope, greed and fear, a rational being might not take the risks that drive good as well as bad aspects of human society.
Marcus points out the ambiguities in human language as a sign of the awkward results of evolution assembling a speech system from older parts. Here, what Marcus sees as a bug, Hawkins might see as a feature or at least a side effect. Ambiguity in language is a result of the generalizing, pattern-matching engine that drives human intelligence. The same design that makes it hard for us to remember details makes it possible for us to recognize and create new patterns.
Marcus’ book is flawed because he compares workings of the human brain with a straw man that has perfect reason. The interesting thing is not how the human mind is perfect, or how it breaks, but why it works the way it does, and how the way things break shows the way things work most of the time.
This is the focus of Oliver Sachs’ Musicophilia. Where Hawkins and Marcus are theorists, Sachs is an anecdotalist. He tells story after story of individuals who gained heightened musical abilities, or diminished musical abilities, due to changes in the brain. The stories themselves are fascinating but have little theory to explain why most humans are attracted to music, or to explain the aspects of the brain that govern parts of musical appreciation and skill.
Reading Musicophilia alongside Hawkins, it makes sense that music plays on the human mind’s attraction to structure and mild surprise. The recent advances in therapies depending on neuroplasticitity are also nicely explained with a theory of how the neocortex is designed for continuous learning. The stories of how memory of musical performance and sequence is retained in patients who can’t store new memories can be explained by a theory that old memories and learned processes are storied differently than new stuff.
Summary recommendations, for Peterme.
* I liked On Intelligence. Some reviewers give Hawkins grief because he brings very little evidence of the neurological or biohistorical basis of his speculations. He was frustrated by the lack of academic support for the kind of science he wanted to do as a younger man; and so he thumbs his nose at the establishment and its puny traditions of supporting arguments. Nevertheless, his argument is testable, so evidence will win in the end. The bigger weakness is his discounting of the role of emotion in intelligence.
* I didn’t like Kluge. I thought it was much weaker than Marcus’ earlier books, which had a stronger grounding in scientific detail, melding infant development, evolutionary developmental biology, and computer modeling. In his earlier books, Marcus built interesting arguments about brain development from rich evidence. This book has some interesting anecdotes, but is mostly a polemic against some common fallacies. The fact that the straw men are common doesn’t make debunking them more interesting; the book reads like he is arguing against poorly educated undergrads.
* I liked Musicophilia, despite its limitations. The book consists of anecdote after anecdote, without much connecting theory; but the stories are interesting, and it’s an entree onto some hopefully more robust studies on music, mind, and neuroplasticity.