Content uploaded by Jordan B Peterson
Author content
All content in this area was uploaded by Jordan B Peterson
Content may be subject to copyright.
Content uploaded by Azim Shariff
Author content
All content in this area was uploaded by Azim Shariff
Content may be subject to copyright.
JB[v.20020404] Prn:14/09/2004; 14:28 F: Z13109.tex / p.1 (41-113)
Anticipatory consciousness, Libet’s veto and
a close-enough theory of free will
Azim F. Shariff and Jordan B. Peterson
University of Toronto
Benjamin Libet concluded that, given the constraints caused by the timing of
consciousness, all action is generated unconsciously. The power of
consciousness, according to Libet, is in selection, in having the ability to veto
each action before it is run to completion. In this paper we challenge the veto,
by proposing that the role of consciousness exists prior rather than after the
initiation of action. This is done through a modulation of the likelihood of
automatic responses to constant stimuli. To account for the time lag of
consciousness, we also suggest that conscious attention is focused on a
predicted future in order to remain effectual to real time existence.
Implications for free will are discussed.
Keywords: Free Will, Conscious Volition, Benjamin Libet, Time-Lag,
Backwards Referral, Supervisory Attentional System
Introduction
Benjamin Libet’s experiments on the timing of conscious awareness have stim-
ulated a long-standing debate regarding the existence and nature of conscious
efficacy. His most heavily cited experiment (Libet 1985) aimed to determine
where in the sequence of a voluntary action conscious awareness enters and
acts. Libet demonstrated that awareness of an action (in this specific case, a
wrist flick) occurs 250–350 msec after EEG-measured reaction potential tech-
nology indicates its onset or preparation. Libet logically concluded that since
consciousness arises after the initiation of an action, it could not be causally in-
volved in the action: “The initiation of the freely voluntary act appears to begin
in the brain unconsciously, well before the person consciously knows he wants
to act!” (Libet 1999:51). If consciousness is acting at all, therefore, it must be
doing so at least 350 msec before such voluntary acts occur.
JB[v.20020404] Prn:14/09/2004; 14:28 F: Z13109.tex / p.2 (113-152)
Azim F. Shariff and Jordan B. Peterson
Further experiments by Libet and others (for instance, Libet, Alberts,
Wright & Feinstein 1967; Keller & Heckhausen 1990) have supported the
concept of a conscious time-lag that would inhibit the ability of consciously-
derived actions to respond in real time. Norretranders (1991) has argued, for
example, that the time lag is a necessary consequence of the immense amount
of stimulus information that must first be processed and then discarded to pro-
vide a coherent and concise story that we can consciously grasp. These conclu-
sions appear logically sound. However, their implications still seem difficult to
accept. If we cannot respond in real time with consciously determined actions,
then our actions are necessarily unconsciously determined. Our perception of
conscious self-efficacy – conscious free will – is illusory.
The hypothesis of free will as an illusion therefore appears to inexorably
relegate consciousness to the realm of epiphenomenalism. Without any force
of its own, consciousness is merely a parallel phenomenon that occurs along-
side our actions – a spectator, not a participant. John Searle (2000) has sug-
gested, however, that the epiphenomenalist explanation of consciousness is
unsatisfying – in part because of its incompatibility with what we understand
about evolutionary processes. First, consciousness is a complex and metabol-
ically expensive biological process, as revealed by demonstration of the differ-
ences in the glucose consumption of unconscious and conscious brain states
(Nofzinger, Mintun, Wiseman, Kupfer & Moore 1997). Second, the rapid in-
crease in human brain size likely contributed to emergence of consciousness in
humans. This encephalization process has resulted in significant evolutionary
costs in the form of danger to both baby and mother during childbirth (Tra-
vathan 1987) and the prolongation of infantile dependence. The adaptive value
of consciousness would therefore logically need to outweigh the costs that ac-
company it. But how much adaptive value could a system that is unable to
make any changes to the actions of the organism have? As far as survival or re-
production is concerned, an epiphenomenal consciousness must be completely
ineffectual. So why incur the costs? Either the epiphenomenalist argument is
wrong, evolutionary theory is wrong, or something is missing entirely. Thus
the problem remains unsolved and is a matter of heated debate.
Libet himself tried to fit the possibility of an active consciousness within
his theory, by arguing that the purpose of consciousness was its ability to veto
any unconsciously derived action. Various members of Libet’s group (for ex-
ample, Libet, Gleason, Wright & Pearl 1983) have shown that it takes 50 msec
to deliver the synaptic message to move one’s wrist. Conscious awareness of
an action appears to emerge about 200 msec prior to the action. This means
that only 150 msec remain for the subject to consciously choose whether or
JB[v.20020404] Prn:14/09/2004; 14:28 F: Z13109.tex / p.3 (152-197)
A close-enough theory of free will
not to proceed with the action. According to Libet’s hypothesis, consciousness
is a filter, capable of selecting which unconscious decisions result in action.
This veto argument, although influential, is not without its problems. Pri-
marily: shouldn’t this conscious choice, concerning veto, be unconsciously
initiated as well? Libet (1999) tries valiantly to indicate that this is not neces-
sary. However, he runs into the following problem: the veto choice is conscious
and immediate, and must therefore occur in a window of 100 msec. Other
decisions, as described previously, cannot be conscious because 300 msec are
required for awareness of choice. In order to reconcile these seemingly incom-
patible ideas, he therefore constructs a complex logical case, describing the veto
as “a control function, different from simply becoming aware of the wish to
act” (p. 53). Finding no direct empirical support for this position, he is forced
in the end to rely on a lack of counter-evidence: “And, there is no experimental
evidence against the possibility that the control process may appear without
development by prior unconscious processes” (p. 53).
Instead of a conscious intervention that affects the action outcome after it
has been initiated, however, it appears possible that the active role of conscious-
ness comes beforehand. In accordance with the Norman and Shallice (1986)
theory of Supervisory Attention, we therefore propose that consciousness acts
in an indirect and more temporally distal role, anticipating and perceiving
upcoming stimuli (more than 400 ms into the future), and modulating the
subject’s largely unconscious preparatory responses. Norman and Shallice’s
model, our contributions to it, and a reinterpretation of Libet’s veto will be
subsequently examined in more detail.
The routinization of behaviour
Shallice has provided an information-processing model of attention, based
on a distinction between automatic and deliberate action that goes back at
least as far as James (1890). In Shallice’s view, automatic and deliberate ac-
tions are handled by two separate systems: the Contention Scheduling System
and the Supervisory Attentional System. Routine, automatic, well-learned be-
haviours that do not require significant attentional resources are handled by
the former system, while novel and non-routine behaviours are facilitated by
the latter. The notion of Contention Scheduling is predicated on the assump-
tion that complex motor programs (schemas, in Shallice’s terminology) can
be unconsciously initiated. This assumption is also shared by Libet, who holds
JB[v.20020404] Prn:14/09/2004; 14:28 F: Z13109.tex / p.4 (197-246)
Azim F. Shariff and Jordan B. Peterson
that all action is initiated without conscious involvement. Direct evidence for
unconscious action initiation has come from many sources.
Australian physiologists Taylor and McCloskey (1990, 1996) have used
backwards masking, for example, to elicit a complex and fully voluntary motor
program, without triggering conscious attention of the eliciting stimulus. As
the subjects of these experiments were not consciously aware of the stimulus
that triggered their behaviour, so the argument proceeds, they could not have
consciously initiated their actions. The validity of this argument is predicated
on the assumption that the motor program elicited by the trigger must have
existed prior to its initiation, as the processes underlying its construction are
so complex that they could not occur (1) without conscious mediation and (2)
in the interval between stimulation and action.
In a follow-up experiment (Taylor & McCloskey 1996), the researchers
showed that subjects can manifest two pre-programmed (previously learned)
voluntary motor responses to two different masked stimuli. Taylor & Mcloskey
have not yet established an upper limit to the number of pre-programmed re-
sponse patterns that might be held in readiness, nor determined how complex
these programs can be. However, the authors do suggest a primarily uncon-
scious and pre-programmed interaction with the world, in a manner that
would allow for Libet’s time-lag of consciousness,
Almost all motor reactions and many other motor performances must occur
before conscious perception of their triggering stimulus. Such a model could
allow the stimulus to act as a trigger when only a small amount of sensory
data has been processed, although conscious perception would require further
sensory input. (Taylor & McCloskey 1990:445)
Recently, Ann Graybiel (1998) has suggested that many motor actions may be
grouped – or chunked – together to form structures similar to the Miller’s
(1957) compression of discrete bits of information in memory) recodes men-
tal representations of action sequences by reducing the number of distinctively
represented units – essentially collapsing several motor programs into one.
Thus, these action repertoires, once elicited, run both automatically and bal-
listically. In Shallice’s contention scheduling, an action schema competes with
others and, when exceeding a threshold, is activated. To avoid conflicting action
patterns, activated schemas laterally inhibit the related and competing ones
(see Cooper, Shallice, & Farringdon 1995 for a discussion and simulation of
the contention scheduling system). Recent work reviewed by Swanson (2000)
lays a plausible hierarchically-structured neurophysiological platform under-
neath these speculations: some sensory input is relayed directly to hypotha-
JB[v.20020404] Prn:14/09/2004; 14:28 F: Z13109.tex / p.5 (246-299)
A close-enough theory of free will
lamic and lower spinal motor control systems, operating very rapidly but at
an “unconscious” level. Another sensory branch relays incoming information
up into cortical areas concerned primarily with complex but time-consuming
high-resolution perception.
However, in non-routine or novel situations, in situations where deliber-
ate attention is required, or when existing schema become error prone, an-
other mechanism is necessary. The Supervisory Attentional System (Norman
& Shallice 1986) does not directly select actions, but applies extra activation
or inhibition to the Contention Scheduling system in order to bias the selec-
tion of particular schemas. Thus, attention directed towards a certain plan
of action greatly increases the likelihood of that plan being selected by Con-
tention Scheduling. For example, while at bat in baseball, one’s attentional
resources are directed towards hitting the oncoming pitch, rather than, say,
catching it. The simple conscious concentration on this goal biases the selec-
tion of the “hit the ball” schema which itself is a hierarchy of efferent muscle
commands controlling the aim, step and swing. In the case of schema mal-
function, when one of the automatic sequences is identified as faulty, it cannot
be fixed in real time, partly for reasons of time constraint, and partly because
the individual components of the sequence have been chunked into a whole
and are no longer available to voluntary regulation at the level of detail re-
quired. The program must therefore be brought into attention and unpacked
into its constituent parts for it to be repaired. The processes underlying playing
a musical instrument, such as a guitar, provide a relevant example. Individual
muscle movements are chunked into the fingering and plucking or strumming
of strings. These notes and chords are themselves chunked into progressions.
The underlying motor “macros” operate ballistically, increasingly, as expertise
with the musical piece is developed. Once a bad habit is established, the attempt
to modify it requires the unpacking, or unchunking, of that part of the hierar-
chy that is now underneath a conscious level. What this means, in essence, is
that “consciousness” moves up a motor hierarchy, as each level of that hierar-
chy becomes automated. Once automated, however, the level is “unconscious,”
and has to be unpacked before it can be repaired or changed.
Indirect conscious control through the biasing of attentional resources
There is strong clinical evidence that the conscious willing of action and the ac-
tual execution of the actions are handled separately within the brain, stemming
primarily from analysis of “double dissociation” disorders – where one process
JB[v.20020404] Prn:14/09/2004; 14:28 F: Z13109.tex / p.6 (299-338)
Azim F. Shariff and Jordan B. Peterson
is damaged but not the other (see Appendix). Some of this clinical evidence
suggests that automatic behavioural macros extend beyond motor sequences,
into the domain of “object” perception – something very much predicated on
use, rather than material or objective feature (Gibson 1979). Indeed, Shallice
and his colleagues (Cooper, Shallice, & Farringdon 1995), describe the pro-
cess of “environmental activation.” Though the presence of a single object or
“stimulus” can initiate a specific schema, they explain, it is more likely that
complex and sophisticated schemas will be activated by more complex, evoca-
tive environmental situations such as combinations or arrays of objects (for
example, a lit match and a dangling cigarette) (see Jeannerod 1997 for a similar
discussion).
However, the idea that such automated perception-motor spanning pro-
cesses exist can also be integrated productively with the idea of the Supervisory
Attentional System. It has been clearly demonstrated that selective attention
can alter the way visual stimuli are perceived, from the simple and meticu-
lously studied reversibility of the Necker cube (Orbach, Ehrlich, & Haith 1963)
to the complex ability to discriminate “objects” from the surrounding sensory
clutter. Consider the well-known cocktail party effect (Cherry 1953) or, more
recently, the results of fMRI studies conducted on the nature of visual atten-
tion by Kastner, De Weerd, Desimone, & Ungerleider (1998). These researchers
discovered that mental representations of various stimuli in a cluttered array
interact in competitive and mutually inhibitory ways. Each stimulus expresses
a suppressive effect on other neighbouring stimuli in the visual field. However,
focusing attention on one particular stimulus offsets the competing suppres-
sive effects induced by these neighbouring stimuli. This mechanism of selective
attention is loosely analogous to the lateral inhibition involved in Shallice’s
contention scheduling. (Cooper, Shallice, & Farringdon 1995). Thus, the Su-
pervisory Attentional System may also be directed at the sensory field, biasing
certain perceptions, rather than their associated motor responses.
One might notice a similarity between the properties that serve as action
cues and the concept of affordances characteristic of J. J. Gibson’s theory of
direct perception (Gibson 1979). According to Gibson, an affordance is what
an object in the environment offers an organism, what it invites from the
perceiver – a chair, for instance, would have the affordance of sitting, a glass
of water for drinking, air for walking through, etc. The three fundamental
properties of Gibsonian affordances are
1. An affordance exists relative to the action capabilities of a particular actor.
JB[v.20020404] Prn:14/09/2004; 14:28 F: Z13109.tex / p.7 (338-401)
A close-enough theory of free will
2. The existence of an affordance is independent of the actor’s ability to per-
ceive it.
3. An affordance does not change as the needs and goals of the actor change.
Though affordances are a useful way of thinking about the action cues of-
fered by objects, Gibson’s definition appears too limited – particularly with
regards to his final two points. In keeping with his over-arching theory of
the directness of visual perception, he maintains that affordances are invari-
ant properties of an “object.” However, the direct perception hypothesis, which
centred around an absence of any perceptual processing, has never gathered
mainstream support (see Fodor, & Pylyshyn 1981 for an early criticism). By
looking at affordances as a property of the shared interaction between subject
and object, and considering that the subject’s perception of the object does af-
fect its affordance, it becomes clear that the affordance of an object can change
not only relative to the action capabilities of a particular subject as Gibson sug-
gests, but between subjects with the same capabilities, and, importantly, within
the same subject. There is no reason why a knife could not reveal the affor-
dance of either cutting bread or spreading butter, depending on the current
goals and attentional focus of a subject. Therefore, it is one’s attention that al-
ters the perceptual interpretation of information; for all intents and purposes,
the same sensory information is seen as a different object, with different affor-
dances and, as a result, different action responses. Indeed, what is seen as an
“object” may be better understood as patterns of environmental information
that, over time, have distinguished themselves from the background as pos-
sessing utility. The potential utility of the object evokes attention, because of
its associated with incentive reward (a consequence of the apprehension of the
potential).
Even in the absence of direct sensory information, “objects” and their af-
fordances can be reinterpreted by directing attention to the standing mental
representations held in memory. Identified objects or situational patterns, pre-
viously understood in a particular context with an associated action, can be
mentally reexamined for further implications for behaviour. Coming to in-
terpret things differently allows one to remap their significance and change
the motor actions associated with them for future encounters. This can be
done rapidly, but not instantaneously (ie under 350 msec), as it does require
conscious attention, and conscious attention, as we have discussed, takes time.
We have thus far suggested that conscious volitional control operates indi-
rectly through modulation of the attentional mental environment. In effect,
the current state of conscious intention biases certain perceptual interpre-
JB[v.20020404] Prn:14/09/2004; 14:28 F: Z13109.tex / p.8 (401-436)
Azim F. Shariff and Jordan B. Peterson
tations of external information. There are motor habits associated with any
given perceptual interpretation, that actually constitute part of that interpre-
tation. Once something is seen in a particular way, therefore, the probability
that certain patterns of action will be implemented is increased, as a matter
of course. The phenomenon of utilization behaviour (Lhermitte et al. 1986),
described in the appendix, provides the clearest evidence for this method of
perceptual response. Prefrontally damaged individuals characterized by uti-
lization behaviour react “automatically” with action to the presence of objects,
and cannot inhibit that automatic response. The prefrontal cortex, whose in-
hibitory capacity is well-documented (Fuster 1993) determines which of these
perceptual-motor programs are relevant and appropriate and disinhibts them,
allowing them to “pass through the gate.” Once the decision has been made, all
other perceptual possibilities and competing affordances are laterally inhibited
further in order to avoid conflicts.
Though our capacity to directly and consciously determine which motor
programs are activated may be severely limited, the same cannot be said of our
ability to affect action For example, a glass of water, under circumstances where
one was thirsty, would elicit a liquid container perceptual scheme and a drink-
ing motor program. However, were there a fire, the affordance interpretation
of a glass of water would be one of a “fire extinguisher” thus activating a very
different motor program. Alternatively, if a nearby staple needed to be flattened
or a stack of windblown papers subdued, the perceptual scheme activated by
the pattern that constitutes the glass could easily be “manipulable solid heavy
surface near at hand” and the affordance interpretation “hammer” or “paper-
weight”, respectively. In this manner, the attentional environment, under the
modulation of a goal-directed consciousness, is the determinant of how ob-
jects are perceived, and which motor programs are likely to be initiated. The
conscious subject need only be aware of her goal to align her automatic actions
with it.
1
Is this free will? If by will, one means the ability to will certain per-
ceptions and actions into a high probability of occuring, then yes. If, however,
one is referring to deliberate, direct control of those perceptions and actions,
then the answer is no. However, this ability seems close enough to free will
to qualify. As conscious beings, we do not force our own hands. Instead, we
whisper suggestions, by modulating our intention and perception. This ability
could be seen as one of the primary functions of consciousness – the ability to
interact with the same world and situation in multiple ways via the plethora of
perceptual and action programs afforded by the flexibility of attention.
One problem remains: Given the delay of consciousness, the time lag
necessary for processing, how can it affect real-time activity? Shouldn’t the
JB[v.20020404] Prn:14/09/2004; 14:28 F: Z13109.tex / p.9 (436-483)
A close-enough theory of free will
attentional environment always be delayed? Shouldn’t this bias the organism
towards anachronistic reactions?
A speculatory solution to this problem is presented in the next section.
Conscious attention, detached from and unable to control real-time
responses, focuses on events that will occur within a time frame to which
it can react
In order for us to consciously modulate our future responses, we need to be
able to anticipate that future. Since there is a relatively long delay in the elic-
itation of conscious sensation, and still more time required for consciousness
to modulate any upcoming response, the attention of consciousness must be
directed far enough in the future for it to be effective. This necessity is clearly
shown in a variety of real-life situations. For example, when skiing, our atten-
tion is not directed to the space immediately in front of our boots. It is too late
for us to consciously affect any outcome relevant to that space. Depending on
our speed, therefore, we pay attention to the terrain a few meters ahead of us
because we are much more able to consciously modulate how we will perceive
and respond to that terrain when we reach it. We learn to take into account how
long is required for our conscious responses. At time frames smaller than this
gap, our responses are already fixed and we are impotent to change them; the
unconscious responses run automatically. We thus have the capacity for free
will, but not over short durations. For this reason, our conscious attention is
most frequently fixed on the future.
2
This attentional fixedness on the future is evident in eye-tracking studies
that examine, for example, the gaze of piano players on their musical notation.
According to the tenets of the theory offered here, skilled pianists consciously
observe the musical notation before them to prime and then elicit previously
automatized, complex repertoires of finger movements. After a given action
repertoire has been initiated, however, any changes to that particular “macro”
would be impossible to implement. Thus, attending to information within the
temporal span of that macro would be useless. We would expect that pianists
would attend to the bars of music suitable for priming the next action macro.
Indeed, skilled pianists look between two and five fixations (the length of a
“look”) ahead (Gilman & Underwood 2003; Truit, Clifton, Pollastek & Rayner
1997; Rayner & Pollastek 1997; Goolsby 1994). Furthermore, experienced mu-
sicians look farther ahead than inexperienced ones. This is precisely what we
would expect, given that more experienced players likely develop larger-scale
JB[v.20020404] Prn:14/09/2004; 14:28 F: Z13109.tex / p.10 (483-545)
Azim F. Shariff and Jordan B. Peterson
macros. Similar behaviours, including the skill difference, characterize touch-
typists (Inhoff & Wang 1992; Salthouse 1984)
Beyond the recognition that consciousness simply attends to the future, we
propose that our subjective experience construes this future as the present, at
least for short-duration decisions. In other words, the anticipated near-future
is referred backwards and interpreted as the present. Nijhawan, (1994) demon-
strated, for example, that experimental subjects can be induced to misperceive
a set of moving dots as further ahead of a set of flashing but stationary dots
when the two sets were actually aligned. Subjects typically perceive the mov-
ing dots on a trajectory line 100 msecs ahead of where they actually are. Thus
their perception of the present actually reflects their prediction of the near fu-
ture. Known as the flash-lag effect, this phenomenon has been demonstrated by
Nijhawan and his colleagues in a variety of permutations of the original experi-
ment (eg, Khurana, Watanabe, & Nijhawan 2003). The phenomenal experience
of moving objects is an extrapolation of where the object will be.
Such results make sense, when considered as anticipatory preparation for
action. To account for the time it takes to become aware of useful informa-
tion and to modulate an appropriate reaction, it is necessary to focus attention
on the future. By perceiving the immediate future as the present, we can ac-
tually believe that consciousness is acting in real time on the (misconstrued)
present, when it is actually preparing for the (likely) future (see Figure A). This
is consistent with Libet’s backward referral, which maintains that the present
is interpreted as the past. Libet showed that a skin prick is only consciously
experienced 500 msec later, but then referred back in time to when it actually
occurred, so that it seems like the sensation was consciously experienced just
(20 msec) after the skin was actually pricked. Thus, in order for consciousness
to seem efficacious, despite its time-lag, the entire subjective interpretation of
time is shifted ahead of the actual time.
This is not something either rare or strange. Klein (1999) and Hameroff
(1999) explain, for example, that such mental juggling of temporal states oc-
curs often, and is quite necessary for coordinating sensory information that
arrives at different times from different parts of the body. The brain is actually
quite capable of reorganizing information to create a smooth and consistent
experience. Because of our capacity to time shift, it is perfectly possible for us
to perceive the predicted future as the present. Because we can predict rea-
sonably accurately, this shift generally causes us no trouble, and simplifies
our interactions with the world. However, the cost of the gap between pre-
diction and actuality is vulnerability to error. When the actual present does not
match the immediately predicted “future,” we place ourselves at risk. Our accu-
JB[v.20020404] Prn:14/09/2004; 14:28 F: Z13109.tex / p.11 (545-584)
A close-enough theory of free will
Time of
stimulus
Libet’s
backwards
referral
Past
Awareness
of stimulus
Attended to
time frame
Backwards
referral of
anticipated
future
Present Future
Conscious
time
Real time
Figure A. The conceptual relation between Libet’s Backwards Referral and Anticipa-
tory Consciousness
racy is compromised during situations that are very rapidly and unpredictably
changing. Since our conscious attention is directed towards a future that hasn’t
happened yet, it can only make predictions to the best of its ability. This is why
not all reactions are perfectly accurate. The ballistic motor program selected is
done with dated information spuriously presented as up-to-date.
Implications for Libet’s veto
In this paper, we have borrowed from, expanded upon and questioned parts
of Libet’s original theory of conscious action and free will. We have outlined
mechanisms that allow, hypothetically, for the existence of constrained free will
and the experimental results that would call free will into question. We have,
however, taken issue with Libet’s veto clause, believing that the capacity for veto
is neither necessary for the existence of free will, nor a manifestation of that will
in the manner Libet proposed. Instead, we suggest that the genuine vetoing of
a command is achieved only by a competing and overriding schema activated
by contention schedule (Cooper, Shallice, & Farringdon 1995). Such a schema
would have enough activation strength to displace whatever schema was cur-
rently being run. More reflexive perceptual schemas and motor reflexes, such
as those characterizing startle and freezing, for example, are the most likely to
be strong enough to overcome previously run macros. The neurocircuitry that
mediates such responses is certainly characterized by the structure that would
enable such displacement (Swanson 2000).
JB[v.20020404] Prn:14/09/2004; 14:28 F: Z13109.tex / p.12 (584-620)
Azim F. Shariff and Jordan B. Peterson
Alternatively, our indirect conscious control inhibition of a response. In
Libet’s experiment, such bias would characterize the wrist flick. However, given
the time frame of such shifts of conscious attention, the biasing would have
to occur prior to the initiation of the action, not within the final 100 msec.
Having biased the activation of the inhibitory schema, it would compete with
and successfully thwart the wrist flick schema using the contention scheduling
process described above. This, however, would not be a veto in the way that
Libet described, since consciousness offered its input before, rather than after,
the initiation of the action. To better understand this process, you might raise
your fist and slam it back down on the table. Do it a second time, but this
time, just before it strikes the surface, stop it from happening and hold your
fist an inch above the table. The conscious inhibition did not occur during the
downward action. It couldn’t. There wasn’t time. Instead, the inhibition came
prior to the action. The almost-strike-the-table-schema was biased over the
strike-the-table-schema. Notice that, this time, you didn’t wince.
Conclusion
The easy and hard problems of consciousness are enormously challenging and
will, no doubt, continue to be so for some time. In addressing the question
of conscious volition, we have faced just one of these challenges. The theory
we offer provides a plausible and coherent solution to some of the problems
that have arisen in the debates regarding the existence and nature of conscious
free will.
The concept of anticipatory consciousness, wherein our conscious percep-
tion of the world is shifted forward to take into account the processing time-lag,
introduces a way that a time-intensive consciousness can remain effective in a
world that often progresses faster than the speed of thought. Meanwhile, we
believe that the two major aspects of consciousness – its phenomenology and
volition – are functionally united by postulating phenomenological awareness
as the key ingredient in conscious choice. The voluntary direction of attention
allows us to consciously bias the selection of a host of ballistic automatic action
patterns. Though are actions are not under direct conscious control, the con-
scious input we do have allows for a working theory of some sort of free will,
or at least something close enough.
JB[v.20020404] Prn:14/09/2004; 14:28 F: Z13109.tex / p.13 (620-659)
A close-enough theory of free will
Appendix: The clinical neuroanatomy of volitional and automatic action
The separation of the willing, initiation and execution of motor action has been
well established, mostly from clinical disorders, but also from directed animal
studies and human neuroimaging. Researchers have attempted to compile this
evidence into an anatomical model of the processes involved in such actions. It
should be noted, however, that making definitive causal claims about specific
structures is difficult and often contested. We will begin by examining some
disorders of the basal ganglia.
Those suffering from Parkinson’s disease, characterized primarily by a loss
of dopaminergic neurons in the substantia nigra pars reticulata (SNr), find
themselves unable to perform the actions patterns they intend to. As Spence
and Frith (1999) indicate, such patients often freeze up altogether, knowing
“precisely what action he wants to perform, but unable to initiate it” (p. 20).
Similarly Huntington’s disease, caused by atrophy of the striatum, results in
both unwanted movements as well as impairment in the acquisition of new
motor skills (Hiendel, Butters & Salmon 1988) Both Huntington’s and Parkin-
son’s disease involve damage to different parts of the basal ganglia, which
have been implicated as the key structures involved in the execution, but not
conscious selection, of action schemas (Spence & Frith 1999). Hikosaka and
colleagues (Hikosaka, Miyashita, Miyachi, Sakai & Lu 1998) have tried to test
this supposition experimentally by using a GABA agonist to create reversible
lesions in parts of monkey brains in order to observe the differential effects
on the learning and performance of visuomotor sequences. They found that
the pre-supplementary motor area (pre-SMA) and the caudate in the basal
ganglia are likely involved in learning new sequences while the putamen and
the cerebellar dentate nucleus are involved either in the storage or retrieval of
these sequences. The differential roles of the caudate and putamen help explain
why Huntington’s disease, a disorder of the entire striatum, affects both motor
acquisition and performance.
Human functional neuroimaging in the same study confirmed the role
of the pre-SMA in learning, and added the dorsolateral prefrontal cortex
(DLPFC). As a sequence became more learned and routinized, activation
moved posterior to parietal regions. They concluded that the prefrontal cor-
tex, the pre-SMA and anterior regions of the basal ganglia initiate the learning
of sequence which, once learned and automatic involve the posterior basal
ganglia, parts of the cerebellum and parietal lobe. Matsumoto and colleagues
(Matsumoto, Hanakawa, Maki, Graybiel, & Kimurai 1999) followed up on
this research, but expanded the role of the striatum to both the acquisition
JB[v.20020404] Prn:14/09/2004; 14:28 F: Z13109.tex / p.14 (659-713)
Azim F. Shariff and Jordan B. Peterson
and retrieval of learned motor programs. They implicated the nigrostriatal
dopamine system as central to the entire process of learning, storing and re-
trieving procedural memory.
Frank, Loughery and O’Reilly (2001) build on this cursory model of ac-
tion selection with reference to the “gating” nature of the basal ganglia. The
projections within the basal ganglia, from the striatum to the globus pallidus
internal segment (GPi) or substantia nigra pars reticulata (SNr) are inhibitory
as are those from the GPi/SNr to the thalamus. In the latter case, it is the con-
stant firing of the GPi/SNr neurons that serves to inhibit the thalamic neurons.
Thus, when the striatal neurons fire, they inhibit the constantly firing GPi/SNr
neurons which, in turn, disinhibits the thalamus (Chevalier & Deniau 1990).
Thus, the firing of the striatum has been referred to as “releasing the brakes”
for motor actions. The actions are selected by cortical regions which project to
the striatum which, in conjunction with the rest of the basal ganglia, “releases”
the motor plan associated with it. Consequently, the thalamus projects to the
motor cortex and the plan is run.
Spence and Frith (1999) break down the system into three parts. The
DLPFC, in conjunction with the anterior cingulate cortex, is involved in the
selection and creation action patterns. Both regions were shown to be differen-
tially activated when subjects were asked to pay attention to their actions (Pass-
ingham 1997). Ingvar and Philipson (1977) have noted similar activation when
subjects were told to simply imagine making movements. Incidentally, Shallice
(1988) places his Supervisory Attentional System squarely in the frontal lobes.
The prefrontal system was contrasted with the subcortical system comprising
the basal ganglia and the cerebellum which were more involved in the execu-
tion than the selection of action. Finally, the parietal cortex is responsible for
the storage of the motor programs.
These results are consistent with the distinction between action intention
and action execution. Just as those with Parkinson’s disease suffer an inability
in action execution, but an intact ability action intention, one would expect to
see the opposite pattern in those with frontal damage, but intact basal ganglia.
Though the frontal regions are much more complex, this expectation is con-
firmed by the clinical evidence from at least two frontal lobe disorders. Graybiel
(1998) describes obsessive-compulsive disorder patients who are compelled
to perform action sequences without their explicit intention. Patients suffer-
ing from Utilization Behaviour (Lhermitte, Pillon & Serdaru 1986) have an
overreliance on environmental stimuli for action. For example, they would au-
tomatically drink from a glass of water placed in front of them, or attempt to
use a toilet every time they walked by. Upon exposure to the relevant cues,
JB[v.20020404] Prn:14/09/2004; 14:28 F: Z13109.tex / p.15 (713-772)
A close-enough theory of free will
these patients cannot help from performing the associated motor plan, even
when entirely inappropriate. Lhermitte described the process as a loss of intel-
lectual control manifested by impairment of frontal lobe inhibition, resulting
ultimately in unrestricted release of parietal lobe activities. Norman and Shal-
lice (1986) suggested that this disorder represents a dysfunctional Supervisory
Attentional System (For a more exhaustive account of relevant evidence, refer
to Norman and Shallice 1986 or Shallice 1988).
Notes
. Many terms have been used to describe the collection of motor responses, including be-
havioural macros (Graybiel 1998), scripts (Schank & Abelson 1977), memory organization
packets (MOPs) (Shank 1982) and schemas (Norman & Shallice 1986). These terms can be
considered largely synonymous.
. A series of experiments by John Bargh and his colleagues (Bargh, Gollwitzer, Lee-Chai,
Barndollar, & Trotschel 2001; Chartrand & Bargh 1996) have revealed that even the goal
behind a pattern of activity may remain outside of conscious awareness. Subjects were un-
consciously primed with certain words that affected their goals and consequent strategies
in an ambiguous situation. These results raise an important note: Consciousness is a suffi-
cient but not necessary condition for the occurrence of the attentional biasing. One could
just as easily allow attention to be drawn to things automatically, as a result of the current
motivation. However, volitional power comes from the ability to purposefully direct this
attention, from, for example typing on a computer screen to the paper on my left to the
paper’s affordance to be picked up and read.
. This conscious biasing can be done more distally and broadly in order to prime certain
fast reactions later. This might help explain why mental visualization improves performance
in skilled activities. By running a mental “simulation” of a certain activity, attention is
primed to be directed at the perceptual-motor macros, hastening their unconscious elici-
tation and improving accuracy.
References
Baldo, M. V., & Klein, S. A. (1995). Extrapolation or attention shift? Nature, 378, 565–566.
Bargh, J. A., Gollwitzer, P. M., Lee-Chai, A., Barndollar, K., & Trotschel, R. (2001) The
automated will: Nonconscious activation and pursuit of behavioral goals. Journal of
Personality and Social Psychology, 81, 1014–1027.
Chartrand, T. L., & Bargh, J. A. (1996). Automatic activation of impression infomation
and memorization goals: Nonconscious goal priming reproduces effects of explicit task
instructions. Journal of Personality and Social Psychology, 71, 464–478.
JB[v.20020404] Prn:14/09/2004; 14:28 F: Z13109.tex / p.16 (772-869)
Azim F. Shariff and Jordan B. Peterson
Cherry, E. C. (1953). Some experiments on the recognition of speech, with one and with
two ears. JournaloftheAcousticalSocietyofAmerica,25, 975–979.
Chevalier, G., & Deniau, J. M. (1990). Disinhibition as a basic process in the expression of
striatal functions. Trends in N eurosciences, 13, 277–280.
Cooper, R. P., Shallice, T., & Farringdon, J. (1995). Symbolic and continuous processes in the
automatic selection of actions’ In Hallam, J. (Ed.), Hybrid problems, hybrid solutions:
Frontiers in artificial intelligence and applications (pp. 27–37). Amsterdam: IOS Press.
Fodor, J., & Pylyshyn, Z. (1981). How Direct is Visual Perception?: Some Reflections on
Gibson’s “Ecological Approach”. Cognition, 9, 139–196.
Frank, M. J., Loughry, B., & O’Reilly, R. (2001). Interactions between frontal cortex and basal
ganglia in working memory: A computational model. Cognitive, Affective & Behavioral
Neuroscience, 1, 137–160.
Gibson, J. J. (1979). Ecological approach to visual perception. Boston: Houghton Mifflin.
Gilmann, E., & Underwood, G. (2003). Restricting the field of view to investigate perceptual
spans of pianists. Visual Cognition, 10, 201–232.
Goolsby, T. (1994). Profiles of processing – eye-movements during sight-reading. Music
Perception, 12, 97–123.
Graybiel, A. M. (1998). The basal ganglia and chunking of action repertoires. Neurobiology
of learning and memory, 70, 119–136.
Fuster, J. M. (1980). The prefrontal cortex: Anatomy, physiology, and neuropsychology of the
frontal lobe. New York: Raven Press.
Hameroff, S. R. (1999). The timing of conscious experience – introduction. In Hameroff, S.
R., Kaszniak, A. W., & Chalmers, D. J. (Eds.), Toward a Science of Consciousness III: The
Third Tucson Discussion and Debates. Boston: MIT press.
Heindel, W. C., Butters, N., & Salmon, D. P. (1988). Impaired learning of a motor skill in
patients with Huntington’s disease. Behavioral Neuroscience, 102 (1), 141–147.
Hikosaka, O., Miyashtia, K., Miyachi, S., Sakai, K., & Lu, X. (1998). Differential roles of
the frontal cortex, basal ganglia, and cerebellum in visuomotor sequence learning.
Neurobiology of Learning and Memory, 70, 137–149.
Ingvar, D. H., & Philipson, L. (1977). Distribution of cerebral blood flow in the dominant
hemisphere during motor ideation and motor performance. Annals of Neurology, 2,
230–237.
Inhoff, A. W., & Wang, J. (1992). Encoding of text, manual movement planning and eye-
hand coordination during copy typing. Journal of Experimental Psychology: Human
Perception and Performance, 18, 437–448.
James, W. (1890). The principles of psychology. New York: Henry Holt.
Jeannerod, M. (1997). The cognitive neuroscience of action. Oxford: Blackwell.
Kastner, S., De Weerd, P., Desimone, R., & Ungerleider, L. G. (1998). Mechanisms of directed
attention in the human extrastriate cortex as revealed by functional MRI. Science, 282,
108–111
Keller, I., & Heckhausen, H. (1990). Readiness potentials preceding spontaneous motor acts:
Voluntary vs. involuntary control. Electroencephalography and Clinical Neurophysiology,
76, 351–361.
JB[v.20020404] Prn:14/09/2004; 14:28 F: Z13109.tex / p.17 (869-962)
A close-enough theory of free will
Klein, S. A. (1999). Do apparent temporal anomalies require nonclassical explanation?
InHameroff,S.R.,Kaszniak,A.W.,&Chalmers,D.J.(Eds.),Toward a Scie nce of
Consciousness III: The Third Tucson Discussion and Debates. Boston: MIT press.
Khurana, B., Watanabe, K., & Nijhawan, R. (2003). Flash lag effect: Speeding up to get
ahead? Journal of Vision, 3 (9), 394a.
Lhermitte, F., Pillon, B., & Serdaru, M. (1986). Human autonomy and the frontal lobes.
part 1: Imitation and Utilization behavior: A neuropsychological study of 75 patients.
Annals of Neurology, 19, 326–334.
Libet, B. (1985). Unconscious cerebral initiative and the role of conscious will in voluntary
action. Behavior and Brain Sciences, 8, 529–566.
Libet, B. (1999). Do we have free will? Journal of Consciousness Studies, 6 (8–9), 47–57.
Libet, B., Alberts, W. W., Wright, Jr. E. W. & Feinstein, B. (1967). Responses of human
somatosensory cortex to stimuli below threshold for conscious sensation. Science, 158,
1597–1600.
Libet, B., Gleason, C. A., Wright, E. W., & Pearl, D. K. (1983). Time of unconscious intention
to act in relation to onset of cerebral activity (readiness potential): The unconscious
initiation of a freely voluntary act. Brain, 106, 623–642.
Miller, G. A. (1956). The magical number seven, plus or minus two: some limits on our
capacity for processing information. Psychological Review, 63, 81–97.
Nijhawan, R. (1994). Motion extrapolation in catching. N a ture, 370, 256–257.
Nofzinger, E. A., Mintun, M. A., Wiseman, M. B., Kupfer, D. J. & Moore, R. Y. (1997).
ForebrainactivationinREMsleep:AnFDGPETstudy.Brain Research, 770, 192–201.
Norman, D. A., & Shallice, T. (1986). Attention to action: Willing and automatic control
ofbehavior.InR.J.Davidson,G.E.Schwartz,&D.Shapiro(Eds.),Consciousness and
self-regulation. Vol. 4.NewYork:PlenumPress.
Norrertranders, T. (1990/1998). The user illusion: Cutting consciousness down to size. New
York: Penguin Books.
Orbach, J., Ehrlich, D., & Haith, H. A. (1963). Reversibility of the Necker cube: I. An
examination of the concept of “satiation of orientation”. Perceptual and Motor Skills,
17, 439–458.
Passingham, R. E. (1997). Functional organization of the motor system. In R. S. J.
Frackowiak, C. Mazziotta, K. J. Friston, & D. Frith (Eds.), Human brain function. San
Diego: Academic Press.
Rayner, K., & Pollastek, A. (1997). Eye movements, the hand eye span and the perceptual
span during sight reading of music. Current Directions in Psychological Science, 6, 49–53.
Salthouse, T. A. (1984). Effects of age and skill in typing. Journal of Experimental Psychology:
General, 113, 345–371
Searle, J. R. (2000). Consciousness, free action, and the brain. Journal of Consciousness
Studies, 7 (10), 3–22.
Schank, R. C. (1982). Dynamic memory. Cambridge: Cambridge University Press.
Schank, R. C., & Abelson, R. (1977). Scripts, plans, goals, and understanding. Hillsdale, NJ:
Erlbaum.
Spence, S. A., & Frith, C. D. (1999). Towards a functional anatomy of volition. Journal of
Consciousness Studies, 6, 11–29.
JB[v.20020404] Prn:14/09/2004; 14:28 F: Z13109.tex / p.18 (962-985)
Azim F. Shariff and Jordan B. Peterson
Swanson, L. W. (2000). Cererbral hemisphere regulation of motivated behavior. Br ain
Research, 886, 113–164.
Taylor, J. L., & McCloskey, D. I. (1990). Triggering of preprogrammed movements as
reactions to masked stimuli. Journal of Neurophysiology, 63, 439–446.
Taylor, J. L., & McCloskey, D. I. (1996). Selection of motor responses on the basis of
unperceived stimuli. Experimental Brain Research, 110, 62–66.
Trevathan, W. R. (1987). Human birth: A n evolutionary perspective. New York: Aldine De
Gruyter.
Truitt, F. E., Clifton, C., Pollastek, A., & Rayner, K. (1997). The perceptual span and eye-hand
span in sight-reading music. Visual Cognition, 4, 143–161.
