How Does the Brain Pay Attention to Interesting Sounds?
The Role of Top-down Signals in Shaping the Listening Preferences of Chameleon
Neurons.
Jonathan B. Fritz - ripple@isr.umd.edu
Stephen
V. David - svd@umd.edu
Daniel
Winkowski - winkows@umd.edu
Pingbo Yin - pyin@umd.edu
Shihab A. Shamma
- sas@isr.umd.edu
Institute
for Systems Research
Department
of Electrical and Computer Engineering
University
of Maryland, College Park, MD 20742
Mounya Elhilali
Department
of Electrical and Computer Engineering
Johns
Hopkins University, 3400 N Charles Street
Baltimore,
MD 21218
Popular
version of paper 5pAB2
Presented
Friday afternoon, April 23, 2010
159th
ASA Meeting, Baltimore, MD
One
of the greatest challenges in understanding how our brains function, now being
intensely studied in many laboratories around the world, is discovering the
neural mechanisms underlying attention. We can effortlessly zoom in on one
conversation at a crowded cocktail party or focus our attention on a violin
soloist playing in an orchestra. How do we do it? Over one hundred years ago,
William James, the famous American psychologist wrote in his Principles of
Psychology (1890): Everyone knows what attention is. It is the taking
possession by the mind, in clear and vivid form, of one out of what may seem
several simultaneously possible objects or trains of thought. Focalization,
concentration, of consciousness are of its essence. It implies withdrawal from
some things in order to deal effectively with others and is a condition which
has a real opposite in the confused, dazed, scatterbrained state of distraction
in the blooming, buzzing confusion of the world.
In
order to make progress in the study of attention, we have recently begun
working on the neural basis of attention in the ferret, a smart and inquisitive
carnivore. Our first insight into the role of attention in modulating brain
function was the discovery that individual neurons in the auditory cortex of
the ferret could swiftly change their listening preferences or receptive
field properties, depending upon what the ferret was most interested in
listening to at that moment. By developing techniques to take snapshot
pictures of neurons dynamically changing listening preferences in the awake,
behaving ferret, we were able to monitor moment-by-moment changes in neuronal
receptive fields as the ferret switched from one attending to one sound or
another for example, changing from focusing on a low tone to focusing on a
high tone, or discriminating a harmonic chord from a noise. Unlike the
traditional view of the neuron, which described a fixed and unchanging
receptive field for each neuronal cell, we soon realized that many neurons have
chameleon like properties that allow them to rapidly change their listening
preferences in order to optimize their abilities to perceive an attended
salient sound. Our results add to a growing chorus of scientific evidence for
extraordinary neuronal plasticity, even in the adult animal, that is mediated
by attention. Similar findings of rapid, adaptive brain plasticity during task
switching and attentional focus have been shown by
brain imaging techniques in humans.
So,
how do chameleon neurons know how to change their listening preferences? Do
they receive an instructive signal from elsewhere in the brain? In order to
answer this question, we began to study the interaction between the frontal
cortex in the ferret, and the auditory cortex. Previous studies in monkeys had
suggested the importance of top-down signals from frontal cortex, and so we
began recording from individual cells in the ferret frontal cortex, while
simultaneously recording from cells in auditory cortex, in order to eavesdrop
on the conversation between these two brain areas during attentive
behavior.
Our
second discovery was that some frontal cortex neurons showed rapid aha! or recognition
responses, which allowed the frontal cells to zero in on the sound of interest
and categorically distinguish between acoustic foreground and background
stimuli thus acting like pure attention cells. The frontal cells did not
respond at all to a sound that was presented in the background. But if the same
sound became the focus of attentive interest, the frontal cells began to
respond dramatically. This neuronal behavior is quite unlike the responses of
neurons in the ferret auditory cortex, which are specifically attuned to
specific acoustic features of the sound. In contrast, the responses of the
frontal neurons were much more abstract and encoded the meaning of the sound,
often independent of the acoustic properties of the sound and were ONLY
interested when the sound was selectively attended.
Our
third insight into the attention system came from simultaneous recordings in
frontal and auditory cortex. We found that there was a striking change in the
coherence of ensemble neuronal activity in the two areas, when the ferret
engaged in behavior, and that this change in coherence was highly specific to
the pitch of the sound that the ferret was attending to. These results suggest
that there is a sharply tuned interaction between frontal cortex and auditory
cortex, in which frontal cortex modulates the specific areas in auditory cortex
that respond to a sound of interest by shining an attentional
spotlight there.
What
would happen to the chameleon cells in auditory cortex if you were to
simulate attentional effects by artificial
stimulation of frontal cortex? In recent experiments, we have stimulated
frontal cortex while simultaneously playing a tone. We find that we can induce
the chameleon cells to change their receptive field pattern in a similar way
that we observe during natural behavior.
These
results suggest that there is an attentional network
for active listening that includes key, interactive components in the frontal
and auditory cortex. Our current research focuses on elucidating other
components in this brain network, figuring out the mechanisms of their action,
and discovering how they work together to allow us to selectively extract and listen
to one voice in a complex world of multiple, overlapping sounds.