Perception is a kind of inference, but it is very different from the kind familiar from our use of
language. When we speak or write, we encounter one sentence at a time, and seem to infer the next
sentence from the ones that came before, just as with a proof in mathematics. But inference in the
brain does not operate in this serial, step-by-step way. Each neuron is connected synaptically with
thousands of others, so its firing pattern is affected by all the neurons that excite or inhibit it, and it in
turn affects the firing of all the neurons that it excites or inhibits. Thus inference is parallel, in that
many neurons are firing at around the same time, and asynchronous, in that there is no central clock
that coordinates the waves of firing that spread through the neural populations. Hence perception is
very different from the kind of serial steps of linguistic inference that have served as models of
reasoning since Aristotle identified such syllogisms as A is B, and B is C, so A is C. Because brains
perform inferences using parallel activity of millions of neurons, perception can elegantly integrate
both bottom-up and top-down information. We will see in chapter 5 that emotional feelings involve a
similar kind of dynamic integration of multiple kinds of information.
Our sense of smell also requires a combination of bottom-up and top down processing. When you
inhale, odorants excite subsets of the millions of cellular receptors in your nose, sending patterns of
neural activity to the olfactory bulb on the bottom of your brain. The olfactory bulb also receives
inputs from other brain areas, such as the hippocampus and the neocortex, so that the signals it sends
to the rest of the brain are already a combination of bottom-up sensory information and top-down
processing. Hence when you smell something, like a live duck in a barnyard or a cooked one in a
restaurant, the smell is the result of dynamic interactions of different brain areas involving both
sensory inputs and previous knowledge and expectations.
The complexity of perceptual processing in the brain shows the implausibility of the traditional
empiricist view that our sense experiences are copies of objects in the world. Without previously
acquired or inherited concepts, we would have a very difficult time dealing with the vast number of
sensory signals that our eyes, ears, and other sensors send to our brains. Perception requires brains to
be able to relate inputs from sensory organs with information they have already stored in the form of
synaptic connections between neurons. Ambiguous examples like the duck-rabbit show that
perception is not just the bottom-up processing of sensory inputs; it also involves top-down
interpretation based on what is already known. Because the brain is a parallel processor capable of
assessing many aspects simultaneously, we do not have to choose between hypotheses that perception
is primarily driven by input to sensory receptors or that it is primarily driven by top-down
interpretation. Rather, brains can perform inferences that simultaneously use both kinds of
information.
Perception is a kind of inference, but it is very different from the kind familiar from our use oflanguage. When we speak or write, we encounter one sentence at a time, and seem to infer the nextsentence from the ones that came before, just as with a proof in mathematics. But inference in thebrain does not operate in this serial, step-by-step way. Each neuron is connected synaptically withthousands of others, so its firing pattern is affected by all the neurons that excite or inhibit it, and it inturn affects the firing of all the neurons that it excites or inhibits. Thus inference is parallel, in thatmany neurons are firing at around the same time, and asynchronous, in that there is no central clockthat coordinates the waves of firing that spread through the neural populations. Hence perception isvery different from the kind of serial steps of linguistic inference that have served as models ofreasoning since Aristotle identified such syllogisms as A is B, and B is C, so A is C. Because brainsperform inferences using parallel activity of millions of neurons, perception can elegantly integrateboth bottom-up and top-down information. We will see in chapter 5 that emotional feelings involve asimilar kind of dynamic integration of multiple kinds of information.Our sense of smell also requires a combination of bottom-up and top down processing. When youinhale, odorants excite subsets of the millions of cellular receptors in your nose, sending patterns ofneural activity to the olfactory bulb on the bottom of your brain. The olfactory bulb also receives
inputs from other brain areas, such as the hippocampus and the neocortex, so that the signals it sends
to the rest of the brain are already a combination of bottom-up sensory information and top-down
processing. Hence when you smell something, like a live duck in a barnyard or a cooked one in a
restaurant, the smell is the result of dynamic interactions of different brain areas involving both
sensory inputs and previous knowledge and expectations.
The complexity of perceptual processing in the brain shows the implausibility of the traditional
empiricist view that our sense experiences are copies of objects in the world. Without previously
acquired or inherited concepts, we would have a very difficult time dealing with the vast number of
sensory signals that our eyes, ears, and other sensors send to our brains. Perception requires brains to
be able to relate inputs from sensory organs with information they have already stored in the form of
synaptic connections between neurons. Ambiguous examples like the duck-rabbit show that
perception is not just the bottom-up processing of sensory inputs; it also involves top-down
interpretation based on what is already known. Because the brain is a parallel processor capable of
assessing many aspects simultaneously, we do not have to choose between hypotheses that perception
is primarily driven by input to sensory receptors or that it is primarily driven by top-down
interpretation. Rather, brains can perform inferences that simultaneously use both kinds of
information.
การแปล กรุณารอสักครู่..