Sense of Self
Monthly Podcast On Cognitive Science
Rhythms: The Oscillations That Bind
Coming Out of the Electrical Synapse
Many factors point to
a functional role for neuronal synchronization.
The speed of our perceptual systems,
which function in real time and require a binding mechanism
that can operate very quickly, argues in favour of some form
of temporal integration such as the synchronization of the
Another argument in favour of synchronization
comes from the well known phenomenon whereby two neurons
that stimulate a third will have a greater effect on it if
their action potentials reach it at the same time.
Consequently, neurons that fire synchronously will tend to
have greater “visibility” and greater associative
Lastly, the neurons in many areas of
the brain already display spontaneous oscillatory activity.
Co-ordinating their activity so that it is in phase (that
is, so that the peaks in their activity are synchronized
with one another) would be an efficient, economical way to
generate synchronization across large distances in the cortex.
Samir Zeki and
his colleagues have shown that even though the various attributes
of a visual scene are presented simultaneously, they are not
perceived at exactly the same time. Colour is perceived
before the orientation of lines, which in turn is perceived
before movement. The lag between the perception of colour
and the perception of movement is on the order of 60 to 80
Zeki’s experiments on visual
perception of the various properties of an object led him
to suggest that an individual’s consciousness may in
fact be composed of a number of micro-consciousnesses corresponding
to the various levels of processing in the brain. Information
would then remain unconscious until it reached a "node" in
the system that rendered it conscious.
Thus a human being’s consciousness
would be composed of a multitude of micro- consciousnesses
that were ultimately integrated into a single, more global
means of language. According to its proponents, this conception
of multiple consciousnesses would allow consciousness to
be understood as something that is truly decentralized, without
any one centre
in the brain from which consciousness springs.
But other authors have objected that the very idea that each
of the subsystems contains a sort of "finish line" where
the stimulus suddenly becomes conscious takes us directly
back to the idea of the Cartesian theatre, which has been
so roundly condemned by philosophers such as Daniel
|NEURONAL ASSEMBLIES AND SYNCHRONIZATION
OF BRAIN ACTIVITY
problem asks the question: how can we have a conscious,
coherent, unified perception of an object, given that its various
attributes are processed in different parts of the brain?
One approach to an answer would be
to say that the various signals from all these different parts
of the brain converge on a group of cells or even a single cell
that represents this conscious perception. Indeed, though the
brain has a preponderance of parallel circuits, it does display
a certain form of convergence
as one moves farther from the primary areas of the cortex and
closer to the “associative”ones. Many such areas
of convergence have been identified in the frontal cortex, the
anterior temporal cortex, and the inferior
parietal cortex. But neuroscientists agree that these are
not sites where mental representations are stored. At most, they
may contain certain “codes”that can reconstruct the
fragments of activations distributed elsewhere in the sensorimotor
areas of the cortex.
Likewise, it is true that there are
certain neurons located at the top of the hierarchy in the visual
cortex that respond specifically to faces, and even to faces
seen from a certain angle. These neurons definitely help us to
recognize faces, because when these neurons are destroyed by
a stroke, the stroke victim develops prosopagnosia,
the inability to recognize faces, even of immediate family.
But apart from a few special cases
such as recognizing faces, which has always been of tremendous
adaptive importance, convergence is too flimsy, inefficient,
and ultimately ineffective a mechanism for unifying perceptions
in the brain, and it does not seem to have been selected by evolution
for this purpose (follow the link to the left).
So if there is no single place where
all of the information about an object converges to become conscious,
is there perhaps a single time when it does? That is the other
major approach to solving the binding problem, and it seems the
more promising. In this approach, broadly speaking, neurons that
are active at the same time are believed to be “perceiving
the same thing”. In technical terms, this approach is based
on the temporal synchronization of neuronal
(After Francis Crick, 1994)
der Malsburg was one of the pioneers of this approach.
In the early 1980s, he began to explore the hypothesis
that the key to the binding problem might lie in the synchronized
activity of the neurons that process the various properties
of an object.
Andreas Engel and Wolf Singer subsequently
confirmed that this hypothesis was well founded. Several
of their experiments seemed to indicate that the objects
represented in the visual cortex are in fact represented
by assemblies of neurons that are firing simultaneously.
To return to the example of the green
suitcase next to the blue hat, each of these two objects
will be represented by a huge assembly of neurons in the brain.
Each of these assemblies will include neurons that can detect
various attributes of each object, such as colour or movement,
or the orientation of its lines and contours, and so on. And
thus, according to this hypothesis, it is through the synchronization
of the various neurons coding for the various attributes of the
suitcase (N1, N2, and N3 in the figure below) that the brain
obtains a coherent, unified image of this object.
As for the neurons that code for the various
attributes of the hat (N4, N5, and N6), they too fire synchronously
to provide a unified image of this object. But as the grey dotted
line shows, they do not fire at the same time as the neurons that
code for the various attributes of the suitcase. And that may be
how we consciously perceive two separate objects standing out against
a background that is also a separate object (and that would be
represented by a third assembly of neurons), and not simply an
undifferentiated amalgam of lines and colours.
Crick and Christof
Koch took the idea of temporal synchronization
one step further by proposing that this synchronized activity,
when occurring at 35 to 75 Hz (hertz, or cycles per second),
may be the neuronal correlate for conscious visual perception.
Their idea originated with studies done on the visual cortex
of cats in the 1980s. These studies had shown that large numbers
of neurons could fire at the same time at a rate ranging from
approximately 35 to 75 Hz. (Neuronal activity oscillating at
this frequency is commonly referred to as “gamma oscillations” or
simply “40-Hz oscillations”.)
Many subsequent studies both in animals and
in humans showed that this high oscillation frequency of neuronal
activity is closely related to the integration of perceptions,
the construction of coherent representations, and to some processes
of selective attention.
Crick and Koch therefore developed a theory according to which
the key to conscious perception lies not solely in the synchronization
of neuronal activity, but in the synchronization of neuronal activity
oscillating at frequencies in the range of 35 to 75 Hz.
In summary, according to this hypothesis:
- if two neurons are oscillating synchronously
in the gamma frequency range (around 40 Hz), then they are
contributing to the same, conscious representation;
- if two neurons are oscillating synchronously
outside the gamma frequency range, then they are contributing
to the same representation, but it is not a conscious one (for
example, the representation of an object that is in your field
of vision but that you are not paying attention to);
- if two neurons are active but are not
oscillating in regular cycles, or are oscillating in regular
cycles that are not synchronized, then these neurons are representing
attributes that are unbound or that are bound to different representations.
Crick and Koch also believe that these transitory
neuronal assemblies oscillating at about 40 Hz do not
form only in the visual cortex but may also recruit neurons throughout
the cortex. In this case, the colour and shape of an object would
not be the only characteristics associated with it. There would
also be all sorts of other characteristics, such as odour, taste,
emotional associations, and so on, all of them thus helping to
form a complete conscious representation of the object observed.
Here then is an elegant mechanism by which the brain might distinguish,
among all the representations that it has bound, those
that are conscious from those that are not.
Crick and Koch had first proposed, in the
late 1980s, that synchronization of the oscillations around 40
Hz was a sufficient mechanism to ensure the emergence of a conscious
perception. Shortly after the year 2000, they altered their position
somewhat, stating that conscious phenomena seem to arise from a competition
among various “coalitions” of neurons (see
box below) in which the winning coalitions determine the content
of the conscious mind at any given time. And what these authors
now consider a more plausible role for the synchronization of the
40 Hz oscillations is that it might act as the mechanism
by which this competition is resolved, by favouring the
selection of a particular neuronal assembly. This idea of neuronal
assemblies that are competing at all times is also consistent with
the more recent hypothesis of the “dynamic
Engel and Wolf also believe in synchronization,
though they consider it necessary but not sufficient to generate
consciousness. In their view, the information should also enter
into a form of short-term
memory, a suggestion that here refers to something like a global
In short, though the the hypothesis of the
synchronization of oscillations at around 40 Hz undeniably contributes
something to our understanding of consciousness, it is not the final
word on this subject. It has, for example, undergone further interesting
developments, particularly as regards the mechanism
by which the brain selects the representations that will become
conscious from among all of its unconscious
of the most interesting empirical data on neuronal activity
and consciousness come from the experiments of Nikos Logothetis.
By recording the activity of isolated cells in the visual
cortex of macaque monkeys, Logothetis demonstrated that the
activity of certain neurons could be correlated with the
conscious or unconscious nature of a stimulus to which these
primates were exposed.
Logothetis used the same method in
these experiments as was used to identify the various
parts of the cortex that specialize in processing different
aspects of visual stimuli (shape, colour, movement,
etc.). But the discovery that the neurons in a particular
part of an animal’s brain always respond to a particular
property of a stimulus does not tell us whether the animal
is perceiving this aspect of the stimulus consciously. We
also know that many visual properties extracted by the most
primary visual areas do not correspond to any conscious perception,
because these neurons continue to respond even when the animal
is under general anaesthesia and hence unconscious.
What was original about Logothetis’s
experiments was precisely that they enabled this distinction
to be made between stimuli that were simply represented by
neuronal activity and stimuli that were perceived consciously.
To make this distinction, Logothetis used rival
stimuli, that is, stimuli that had mutually
exclusive interpretations—in this case, images that
could be interpreted as depicting either upward movement
or downward movement. With training, the monkeys in these
experiments learned to indicate with their hands which direction
of motion they perceived.
In the MT area of the visual cortex,
which is responsible for detecting movement, Logothetis then
found neurons whose response fluctuated in accordance with
the monkey’s learned behavioural response. For example,
certain neurons responded strongly when the monkey was indicating
that it was perceiving an apparent upward movement. But these
same neurons were far less active when the monkey was indicating
a downward movement.
This discovery was important, because
it clearly established that the activity of neurons in the
sensory areas of the cortex does not always correspond solely
to the properties of an external stimulus. For example, when
we compare the trials in which a given neuron was very active
with those in which it was less active, we see that absolutely
nothing had changed in the external stimulus or in the general
conditions of the experiment. The only difference
that corresponded to the change in the activity of this neuron
was the perceived movement that the monkey reported by
means of a behavioural response that it had learned previously.
Some people might argue that the activity
of these neurons merely accompanied the motor output of this
behavioural response. But the location and connectivity of
these neurons makes this hypothesis difficult to sustain.
This type of experiment on the activity
of isolated neurons thus seems to show a direct effect of
higher-order attention processes on sensory processing centres.
But whether the activity of these neurons in the MT area
neuronal correlate of consciousness of movement
all on its own remains an open question.