Learning
Even snails and slugs can learn. The American neuroscientist Eric Kandel won a Nobel Prize for his
research on how learning works in the sea slug, Aplysia. These slugs have only a few thousand large
neurons, so Kandel was able to map out the connections among them and investigate the chemical
mechanisms responsible for the formations of these connections. Sea slugs don't learn much, but they
are able to modify such behaviors as eating and withdrawing from noxious stimuli. Kandel showed in
the 1960s how these behavioral changes result from changes in synapses, the connections between
neurons. For example, when a sea slug is exposed to a new substance and given an electric shock, its
neurons undergo chemical changes that alter its behavior, enabling it to avoid the substance.
Much later, Kandel was able to show that sea slugs experience Hebbian learning, a process
hypothesized by the Canadian neuroscientist Donald Hebb. This kind of learning is captured in a
slogan that summarizes how two neural connections are formed: what fires together wires together.
Consider two neurons with a weak synaptic connection that are both made to fire by the same
stimulus. According to Hebb, there should be a mechanism by which their firing at the same time
wires the neurons together by increasing strength of the synaptic connection between them. From the
work of Kandel and many other researchers, we now know that this kind of learning occurs in sea
slugs and also in animals with much more complex brains.
Cognitive neuroscience is still far from having a full explanation of all the different kinds of human
learning right up to the most creative leaps made by human scientists, inventors, and artists. But
thanks to research by Kandel and others, there is no doubt that many kinds of learning are the result of
identifiable brain processes. Conditioning and Hebbian learning occur in humans as well as lower
animals. The psychiatrist Norman Doidge has written an accessible book about neuroplasticity, the
enormous adaptability of the human brain that is the result of its learning mechanisms. We do not need
to have a fully worked-out account of every kind of human learning to note the substantial ongoing
explanatory successes of the hypothesis that learning is a brain process.
Inference and Language
Most cognitive science research on inference, problem solving, and language has developed
psychological rather than neurological explanations. But progress is rapidly being made on neural
explanations of high-level thinking, and I will give only a few examples. John Anderson is a
psychologist well known for his computational models of problem solving, and he has increasingly
tied these models to the operations of particular brain areas. Vinod Goel has used brain scanning to
identify neuroanatomical correlates of high-level reasoning. Jerome Feldman has proposed a neural
theory of the learning and application of language.
Reading is a practically important example of inference and language, and a leading researcher on
dyslexia has recently reviewed the current state of knowledge of how brains manage to read.
Maryanne Wolf points out that literacy is a recent development in human history, going back only
about five thousand years, to the Sumerians. There is no evidence that the brain evolved special
functions to support reading; rather, reading is like many other cultural developments in using neural
mechanisms that evolved for other reasons. Wolf describes how successful reading requires
interactions among several brain areas, including occipital, temporal, and frontal regions. Difficulties
in reading can arise because of problems with particular areas, such as the angular gyrus, but also
because of interactions between or among different areas. Neural explanations of reading ability and
dyslexia are still sketchy and provisional, but the prospects for further advances in the understanding
of these and other features of language use appear strong.
In order to have a full neural theory of human thinking, we would need to have explanations of how
the brain carries out the most high-level, creative kinds of thought. Theories are just beginning to be
developed of how the brain manages its most impressive feats, like solving challenging problems,
writing novels, composing music, and creating scientific theories. There are rough ideas about how
the brain manages to be creative, but nothing yet that could count as a mechanistic explanation. How
does the brain form new scientific concepts, such as sound wave, electron, and gene? How can
groups of neurons generate new hypotheses, like the idea that species evolve by natural selection?
More mundanely, how do neurons carry out basic forms of inference such as deduction, generalization
from examples, and analogy? The present shortage of available answers to these questions is not
evidence against the hypothesis that minds are brains. That much remains to be understood about
thunderstorms does not undermine the fact that identification of lightning with atmospheric electrical
discharge has had great explanatory success. Every scientific theory is incomplete in that there are
some relevant phenomena that it cannot explain, but such gaps become evidence against a theory only
when an alternative theory arises that can fill them by explaining the phenomena. The view that minds
are souls cannot explain creativity and high-level inference either, and lacks any prospects for
explanatory progress.