continued, Section 3 of 4
Third Criterion: Subjectivity
the quicksilvered criterion, the most elusive of the three Great Criteria
of personal identity.
The philosopher Albert Shalom,
writing in the mid-80s, could grant corporeal bases for memory and continuity
— but not for subjectivity. In his view subjectivity is
There seems... to be no alternative but
to regard the continuity of body or, according to more recent writings, the
biographical or autobiographical continuity of experiences, as the only
basis on which one can refer to an individual's "identity"....
But when we
are dealing with subjectivities, in particular that of the human entity, the
situation is not quite so simple. It is through our own awareness of
ourselves that this simple model is most clearly seen as inadequate.  ...[T]he
constant ability of an "I" to be aware of "itself"
throughout all bodily changes and throughout all the changes of subjective
experience, is a problem which simply cannot be dealt with in terms of that
central difficulty... of all theories which try to derive an adequate theory
of the person from the body/mind framework [is that] there is no explanation
of how neural activity can give rise to subjective processes at all.
It follows that all such theories can be no more than variations on the
inexplicable emergence of subjectivity from an initial fertilized ovum
considered as a purely physicochemical entity.
That was a mid-80s perspective. For a
mid-90s perspective, we can turn to another accomplished philosopher, David Papineau.
Papineau makes a novel comparison. He likens the twentieth-century study of
consciousness — subjectivity in its fullness — to the nineteenth-century study of life:
It may seem to some readers that a
physicalist theory of consciousness will come close to denying the existence
of consciousness. But that would be a mistake. It doesn't deny
consciousness, just a certain conception of consciousness.
that consciousness is some kind of extra inner light, some further
non-physical property which exists over and above any physicalistically
specifiable property. But this is quite consistent with holding that
consciousness is a real property which distinguishes some kinds of systems
from others. This combination of views requires only that we accept
that consciousness is identical with some property which is specifiable in a
physicalistically acceptable way.
may be helpful here. In the nineteenth century there was a heated
theoretical debate about the essence of life. The participants had a satisfactory enough working
notion of life: they agreed about which kinds of behaviour and
physical organization are characteristic of life, and in consequence were
clear enough about where in practice the line should be drawn.
Everything from humans to microbes is alive, while planets and pebbles are
dead. (Perhaps there were some borderline cases; but the penumbra of
vagueness was not wide.)
despite this wide degree of agreement on the nature of life,
nineteenth-century thinkers took there to be a further question.
Why are these systems alive? What mysterious
power animates them? And why is this power present in certain cases,
such as trees and oysters, and not in others, like volcanoes and clouds?
questions have disappeared from active debate. Biology textbooks
sometimes begin with a few perfunctory paragraphs about the distinguishing
characteristics of their subject matter. But the nature of life is no
longer a topic of serious theoretical controversy. Everybody now
agrees that the difference between living and non-living systems is simply
having a certain kind of physical organization (roughly, we would now say,
the kind of physical organization which fosters survival and
explanation for this nineteenth-century debate, and its subsequent
disappearance, was that it was premised on the notion that living systems
were animated by the presence of a special substance, a vital spirit, or
elan vital, which was postulated to account for
those features of living systems, such as generation and development, which
were thought to be beyond physical explanation. And of course, if you
do believe in such a vital spirit, then you will want to know about its
nature, and why it arises in certain circumstances and not others.
nobody nowadays believes in vital spirits any more, not least because it is
now generally accepted that the characteristic features of living systems
can in principle all be accounted for in physical terms. In
consequence, it no longer makes sense to puzzle about why living systems are alive. To be alive is just to be a
physical system of a certain general kind. There isn't any extra
property present in living systems, over and above their physical features,
which distinguishes them from non-living systems. So we have stopped
asking questions which presuppose such an extra property.
that we do the same with consciousness....
reject the theory that consciousness involves an extra inner light in
addition to facts of physical organization. But we can reject this
theory without rejecting consciousness. Even if consciousness is just
a kind of abstract physical organization, the difference between being
conscious and not being conscious can still be perfectly real.
Papineau predicts positive results from
physicalist studies of consciousness. Two very recent (1997)
studies are especially relevant to this essay, in that they point the way towards a corporeal subjectivity
criterion. The studies isolate subjective mechanisms within their functional models of conscious brain systems.
This functional approach may well address Shalom's
concern, and vindicate Papineau's faith in wrinkled clay. We'll review the studies below.
Bernard Baars has developed a "Global
Workspace" model of consciousness, as described in a joint article by James Newman, Bernard Baars and Sung-Bae Cho,
entitled, "A Neural Global Workspace Model for Conscious Attention."
characterize attention as a "global integration and
dissemination system... which controls the allocation of the processing resources
of the central nervous system."
system model places great importance on the thalamus.
The thalamus is a deep-seated limbic
structure. Like the hippocampus, it also intercepts transmissions along
the brain's sensory pathways. But the thalamus is not an organ of memory
like the hippocampus — instead, the thalamus is known to be
a regulator of "attentional states," or states of alertness.
Central location of
thalamus has connections to and from many
regions of the neocortex.
For Newman, Baars and Cho, the thalamus performs attentional functions as part of the Global
Workspace model. The Global
Workspace is a competitive neural net system wherein
independent neural output units compete with each other to see which unit can
produce the largest response to each environmental stimulus input. The
output unit with the largest response wins the competition. It
proceeds to categorize the stimulus, classifying it as a
distinct feature of the external environment.
In the thalamus,
this competition is thought to "gate" attention. Gating
blocks off unimportant sensations before they can reach the neocortex; freeing
the cortex to act upon just those features of its environment which
are novel, or else important to survival.
Some long-distance connections between the thalamus and the neocortex
alerts the neocortex to these important environmental features. In Figure
8.3 the thalamus is the dark ovoid at the center of the inset image. The larger, main
image displays thalamic fiber bundles — they run from the thalamus to other brain
regions (which reciprocate with fibers of their own). Note
the thickness of the fiber bundles. Like high-voltage power lines, these fibers transmit thalamic
alerts with swift and abundant force.
A "wagon wheel" diagram of central nervous system attention and conscious perception. "Th" is
the thalamus; "gc" is a
closed attention gate; and "go" is an open attention gate.
In this diagram a visual stimulus has opened an attention gate to the primary
visual cortex (V1).
In Figure 8.4 a
"wagon wheel" diagram of the Global Workspace places the thalamus at
visual center. Newman, Baars and Cho speak of it as the "hub" of
their model. Thick fiber "spokes" anchor it to all points on the rim of cortex. In Baars' model the thalamus maintains gated attentional
connections to all regions of the central nervous system.
Workspace model outlines a neural mechanism of attention.
And this is highly relevant to the subjectivity criterion; because
when the mind "pays attention" to some object, it maintains a distinction between itself
and that object. The distinction between self and object can also be stated as the
distinction between self and not-self, which is perhaps a more general
definition of subjectivity. Through this "verbal overlay" we can see more clearly
the model's relevance to the subjectivity criterion.
Newman, Baars and Cho
summarize the most important properties of their model:
[W]e have introduced a collection of
neuroscience and [neural net] models for attention and binding, resource
allocation, and second-order gating, which share important features and
parallels with a neural Global Workspace System for conscious
attention.... While the [neural net] models we have presented only implement
partial aspects of the [Global Workspace] system, and even our Wagon Wheel model largely
neglects the influences of memory and affective systems upon the stream of
consciousness, the outlines of a general framework for understanding
conscious processes should be discernable.... This is certainly great
progress, given the virtual terra incognita
consciousness has been for most of the history of science.
John G. Taylor builds upon this Global
Workspace model. In "Neural Networks for Consciousness"
Taylor presents a
complementary neural net model. His model distinguishes
discrete stages in the emergence of consciousness.
his exposition by listing several features of consciousness which any
detailed model (neural net or other) must mimic. Taylor categorizes these
low-level features of consciousness, as in Table 8.1. Here, "spatial features"
place the location of mental activity; "temporal features" delimit the timing of
mental activity; and "emergent features" specify the ways in which mental activity
emerges as conscious awareness, or "phenomenal experience."
Criterial features of
phenomenal experience (PE)
continuation of activity
required to achieve PE
between different PE
creation of PE activation
created at highest level
emergence once begun
No ability to probe lower levels
supporting PE but below PE
These features of conscious life have been extracted
from a mountain of psychological studies. Taylor
is careful to work within the real-world limitations which they impose upon the
To model the emergence of consciousness
at its lowest level,
Taylor begins with a two-stage neural net model of cortical processing.
The two stages correspond with the
two lowest levels of mental activity: the first stage is pre-conscious, an
"unattended condition"; the stage above it is "passive
awareness" (awareness without attention). The third and highest stage, the "attended
condition," Taylor models separately.
The images below map Taylor's three stages
to three distinct regions of cortical activity. Figure 8.5 is the
predictive theory. Figure 8.6, the experimental result.
Levels of activation in
modules of the three-stage model, and their corresponding levels of
Images of brain activity
(dark areas) corresponding to the three levels of conscious activation.
"A" shows the unattended condition. "B" shows
passive awareness, minus the unattended condition. "C" shows the
attended condition, minus passive awareness and the unattended condition.
The result agrees with prediction: each stage of awareness activates additional cortical regions. To see better why this is the case, we can highlight a broad correspondence between
Taylor's three stages of awareness and three sectors of Baars' "wagon wheel" attentional model.
Figure 8.7 illustrates with color-coded sectors. Note that passive awareness (yellow)
charges out of the thalamus' attention gates. When a gate opens, additional cortical regions are activated.
Correspondence between three stages of awareness and three sectors of the
"wagon wheel" attentional model.
This broader correspondence, which is as much anatomical as theoretical, bolsters Taylor's argument. In Taylor's model, attentional systems gate thoughts between these
discrete stages of awareness. As it happens, Taylor's attentional systems are similar to the
thalamic system theorized by Newman, Baars and Cho (as above). There is in fact
some overlap between the proposed systems, in and around the thalamus.
clinical results in support of his model, along with details of the model's
continuum neural field theory. That mathematical theory defines
recursive "bubbles" of persistent cortical
activity: bubbles set in motion by the thalamus. These recursive bubbles separate stage-two thoughts (passive awareness) from
stage-one thoughts (the unattended condition).
For Taylor, this second-stage recursive separation is the principal requirement of
In the model, the lower first stage involves modules tranforming
inputs into various codes. It is only at the second level that further
modules are activated in a manner which brings about the
emergence of phenomenal awareness.
Awareness, even passive awareness, entails subjectivity — it's the conscious subject, after all, who is aware.
This ubiquitous correlation indicates that the transition
from the unattended condition up to passive
awareness marks the emergence of subjectivity. (It's a popular idea: many philosophers are now willing to entertain just this sort
can be stated more dramatically. We can say that Taylor's third stage of consciousness — the "attended
condition" — is not necessary for
subjectivity. I should clarify this more dramatic statement:
The attended condition engages the brain's frontal lobes for
sustained concentration, and for planning and reasoning tasks which pertain to the object of attended concern. Such effort is a hallmark of the human
experience, but adult humans sustain the attended condition only
intermittently. Moreover, this level of consciousness is entirely absent
in young children. Taylor explains:
[T]he late onset of prefrontal myelination
[axon sheathing] would prevent much use being made of such frontal sites in
the first few years of life, a period in which there is extensive coding of
words and objects in posterior sites.
Taylor goes on to cite studies of
adults who have maintained conscious life despite damage to the frontal
These case studies
provide additional support to the idea that third-stage thoughts can be distinguished
from, and are largely independent of, basic subjectivity.
The implication of Taylor's
work is that subjectivity emerges within the second stage of
mental activity — not the third. Passive awareness would appear to
and Cho reach a similar conclusion. They deduce that the frontal lobes
perform an "executive function" over the limbic system.
The researchers take pains to point out
the fact that this executive function is not
indispensable to consciousness — thalamic competition is judged the more fundamental
manifestation of conscious life:
Note that an executive system is not an essential requirement for consciousness. That
this is not the case is illustrated by the literature on extensive damage to
the frontal lobes of the brain.... In terms of the [Global Workspace] model we have
presented, it is not executive attentional processes, but the selective
binding of coalitions of active cortical modules via a thalamocortical
competition which is the sine qua non for the
generation of a coherent stream of conscious representations.
And finally, a picture worth a thousand words: Figure 8.8 shows the thalamus at work.
Thalamus imaged while in the process of stimulating the cortex to attention.
This extraordinary set of images comes from an experiment
in which five adults were monitored as they focused attention upon a visual target. These images are a record of activity within
each subject's thalamus (blue rectangle).
Subjects attended to the target
under three different mental conditions: high, medium and low arousal. Performance during the state of low arousal
(morning sleep deprivation) ignited the greatest activity in the thalamus. This activity maps to
color pixels, which
predominate in the "low arousal" images at left.
The investigators speculate that this
record of heightened activity captures the thalamus as it is "working harder" to force a drowsy cortex up into an attentive state.
thalamus, it seems, is here compelling the cortex to "wake up and pay attention."
Remarkable images —
and photographic evidence in support of theories which implicate the thalamus in subjective awareness.
So thalamus is king. Taken all in all, the thalamic attentional system would appear to be of utmost importance
to the emergence of subjectivity.
Conclusion: The subjectivity criterion of personal identity has a
next Section 4 of 4
Chapter 8, Section 3 Endnotes
David Papineau, Philosophical
(Oxford: Blackwell Publishers, 1993) 121-23.
James Newman, Bernard J. Baars, and Sung-Bae Cho, "A Neural
Global Workspace Model for Conscious Attention," Neural Networks
10:7 (1997): 1195-1206. The article is reproduced within the next
Newman, Baars, and Cho 1195.
See especially Newman, Baars, and Cho 1197-99, Section 2.
Marvalee H. Wake, ed., Hyman's Comparative Vertebrate Anatomy
, 3rd ed.
(Chicago: University of Chicago Press, 1992) 681. Quoting:
"The fifth division of the dorsal
thalamus is composed of the intralaminar
nuclei. The neurons of this group are scattered along the margins of
the other thalamic nuclei and constitute an extensive
arousal system to the telencephalon. These neurons receive
projections from all of the ascending sensory pathways..., and they project
to the outer layer of the isocortex.
appear to form an alerting or biasing mechanism for the isocortex.
They probably do not convey information on where or what is happening in an
animal's world, but rather convey that something new or different is
happening that 'needs' to be analyzed."
Francis Crick, The Astonishing Hypothesis
(New York: Charles Scribner's Sons, 1994) 84.
For a good introduction to competitive neural networks,
see McLeod, Plunkett, and Rolls 127-38, Chapter 6.
Thick axons are fastest. Thalamocortical fiber bundles are thick enough to be visible to the unaided eye: the lobotomy procedure severs them. For neuro-computational details
of their alerting function, see Anthony Zador, "Thalamocortical Synapses: Sparse but Stentorian," Neuron
23:2 (1999): 198-200.
Newman, Baars, and Cho 1199. Quoting the original
"'Wagon wheel' model of CNS systems
contributing to global attention and conscious perception. A1, primary auditory area; BG,
basal ganglia; gc, 'closed' nRt gate; go, 'open' nRt gate;
MRF, midbrain reticular formation; nRt, nucleus reticularis thalami; PFC, prefrontal cortex; S1, primary somatosensory area; Th, ventral thalamus;
V1, primary visual cortex (from Newman et al., 1997)."
Newman, Baars, and Cho 1205.
John G. Taylor, "Neural Networks for Consciousness," Neural Networks
10:7 (1997): 1207-25. The article is reproduced within the next
John G. Taylor, "Neural Networks for
John G. Taylor, "Neural Networks for
John G. Taylor, "Neural Networks for
John G. Taylor, The Race for
Massachusetts: A Bradford Book, MIT Press, 1999) 270. Quoting the original
"Surface-projected schematic representation of mean
activated areas of cortex (a) during inattention to an auditorily presented
stream of syllables due to attention to a visual signal presented to both
eyes. (b) The additional cortical area brought on stream when
subjects passively listen to the syllables (no visual distractor).
(c) Additional cortical area when attention is focused on the
The correspondence is this author's interpretation.
Note especially that the juxtaposed
images do not represent attentional events of the same sensory modality. The wagon wheel drawing
illustrates a visual stimulus, whereas the brain images record an auditory stimulus.
For the researchers' original captions, refer to notes 28 and 34 of this section.
John G. Taylor, "Neural Networks for
Consciousness" 1209-11, 1221-22.
Taylor's model of the "thalamus-NRT-cortex complex"
is cited by Newman et al. See Newman, Baars, and Cho 1198.
John G. Taylor,
"Neural Networks for
In light of prior arguments, it is perhaps unsurprising to find
that Taylor's hypothesized neural net architecture is massively recursive.
See especially John G. Taylor, "Neural Networks for
Consciousness" 1210, Figure 4; also reproduced within the next
. Taylor provides a supplementary exposition on recursion
in the nucleus reticularis thalami (the layer of neurons surrounding the
thalamus proper) in John G. Taylor, The Race for
John G. Taylor, "Neural Networks for
Consciousness" 1207. (Quoted from the abstract.)
For example, David Chalmers, a prominent and popular philosopher of mind,
has recently (1996) made a strong case for something very similar: namely, the correlation between consciousness
and awareness. See David J. Chalmers, The Conscious Mind
(New York: Oxford University Press, 1996) 213-46; Chapter 6. Quoting from page 243:
"It is very plausible that some kind of awareness is necessary
for consciousness. Certainly all the instances of consciousness that I know
about are accompanied by awareness. There seems to be little reason to believe in any
instances of consciousness without the accompanying functional processes.
If there are any, we have no evidence for them, not even indirect evidence, and we
could not in principle. It therefore is reasonable to suppose on the
grounds of parsimony that wherever there is consciousness, there is awareness."
John G. Taylor, "Neural Networks for
John G. Taylor, "Neural Networks for
Newman, Baars, and Cho 1201.
Newman, Baars, and Cho 1202.
C. M. Portas, G. Rees, A. M. Howseman, O. Josephs, R. Turner, and C. D. Frith,
"A Specific Role for the Thalamus in Mediating the Interaction of Attention and Arousal in Humans,"
The Journal of Neuroscience
18:21 (1998): 8982.