Neuroscience of free will
Encyclopedia
Neuroscience of free will refers to recent neuroscientific investigations shedding light on the question of free will
, which is a philosophical
and scientific
question as to whether, and in what sense, rational agent
s exercise control over their actions or decisions. As it has become possible to study the living brain
, researchers have begun to watch decision-making processes at work. While findings may carry implications for moral responsibility
in general, the field remains highly controversial as there is no consensus among researchers about the significance of findings, their meaning, or what conclusions may be drawn.
" can encapsulate different hypotheses, each of which must be considered in light of existing empirical evidence.
It is quite likely that a large range of cognitive operations are necessary to freely press a button. Research at least suggests that our conscious self does not initiate all behaviours. Instead, the conscious self is somehow alerted to behaviours that the rest of the brain and body are already planning and performing. These findings do not forbid conscious experience from playing some moderating role (although this page does discuss studies suggesting that moderating and cancelling an action is an unconscious process too). The key is that the unconscious processes play a much larger role in behaviour, because they are what first initiate actions, whereas self awareness may only recognize that an action has been prepared.
It may be possible, then, that our intuitions about the role of our conscious "intentions" have led us astray; it may be the case that we have confused correlation with causation
by believing that conscious awareness necessarily causes the body's movement. This possibility is bolstered by findings in neurostimulation
, brain damage
, but also research into introspection illusions
. Such illusions show that humans do not have full access to various internal processes. The discovery that humans possess a determined will would have implications for moral responsibility
. Neuroscientist and author Sam Harris
believes this is the case, but says we should never have expected we had libertarian free will. He argues that "Thoughts simply arise in the brain. What else could they do? The truth about us is even stranger than we may suppose: The illusion of free will is itself an illusion".
There have been a number of problems regarding studies of free will. Particularly in earlier studies, research relies too much on the introspection of the participants (other research shows that things such as introspective estimates of event timing are not accurate). The conscious decision may be made elsewhere in the brain before conscious realization, leaving room for free will even if the conscious realization comes delayed. Many brain activity measures have been insufficient and primitive as there is no good independent brain-function measure of the conscious generation of intentions, choices, or decisions. The conclusions drawn from measurements that have been made are debatable too, as they don't necessarily tell, for example, what a sudden dip in the readings are representing. In other words, the dip might have nothing to do with unconscious decision, since many other mental processes are going on while performing the task. Some of the research mentioned here has gotten more advanced, however, even recording individual neurons in live volunteers.
These studies have only just begun to shed light on the role that consciousness plays in actions and it is too early to draw very strong conclusions about many "free will"s. It is worth noting that such experiments - so far - have dealt only with free will decisions made in short time frames (seconds) and may not have direct bearing on free will decisions made ("thoughtfully") by the subject over the course of many seconds, minutes, hours or longer.
in the 1980s, in which he asked each subject to choose a random moment to flick their wrist while he measured the associated activity in their brain (in particular, the build-up of electrical signal called the readiness potential). Although it was well known that the readiness potential preceded the physical action, Libet asked how the readiness potential corresponded to the felt intention to move. To determine when the subject felt the intention to move, he asked her to watch the second hand of a clock and report its position when she felt that she had felt the conscious will to move.
Libet found that the unconscious brain activity leading up to the conscious decision by the subject to flick his or her wrist began approximately half a second before the subject consciously felt that she had decided to move. Libet's findings suggest that decisions made by a subject are first being made on a subconscious level and only afterward being translated into a "conscious decision", and that the subject's belief that it occurred at the behest of her will was only due to her retrospective perspective on the event.
The interpretation of these findings has been criticized by Daniel Dennett
, who argues that people will have to shift their attention from their intention to the clock, and that this introduces temporal mismatches between the felt experience of will and the perceived position of the clock hand. Consistent with this argument, subsequent studies have shown that the exact numerical value varies depending on attention. Despite the differences in the exact numerical value, however, the main finding has held. Philosopher Alfred Mele
criticizes this design for other reasons. Having attempted the experiment himself, Mele explains that "the awareness of the intention to move" is an ambiguous feeling at best. For this reason he remained skeptical of interpreting the subjects' reported times for comparison with their 'readiness potential'.
In a variation of this task, Haggard and Eimer asked subjects to decide not only when to move their hands, but also to decide which hand to move. In this case, the felt intention correlated much more closely with the "lateralized readiness potential
" (LRP), an ERP
component which measures the difference between left and right hemisphere brain activity. Haggard and Eimer argue that the feeling of conscious will must therefore follow the decision of which hand to move, since the LRP reflects the decision to lift a particular hand.
A more direct test of the relationship between the readiness potential and the "awareness of the intention to move" was conducted by Banks and Isham (2009). In their study, participants performed a variant of the Libet's paradigm in which a delayed tone followed the button press. Subsequently, research participants reported the time of their intention to act (e.g., Libet's "W"). If W were time-locked to the readiness potential, W would remain uninfluenced by any post-action information. However, findings from this study show that W in fact shift systematically with the time of the tone presentation, implicating that W is, at least in part, retrospectively reconstructed rather than pre-determined by the readiness potential.
A study conducted by Jeff Miller and Judy Trevena (2009) suggests that the readiness potential (RP) signal in Libet's experiments doesn't represent a decision to move, but that it's merely a sign that the brain is paying attention. In this experiment the classical Libet experiment was modified by playing an audio tone indicating to volunteers to decide whether to tap a key or not. The researchers found that there was the same RP signal in both cases, regardless of whether or not volunteers actually elected to tap, which suggests that the RP signal doesn't indicate that a decision has been made.
In a second experiment, researchers asked volunteers to decide on the spot whether to use left hand or right to tap the key while monitoring their brain signals, and they found no correlation among the signals and the chosen hand. This criticism has itself been criticized by free-will researcher Patrick Haggard, who mentions literature that distinguishes two different circuits in the brain that lead to action: a "stimulus-response" circuit and a "voluntary" circuit. According to Haggard, researchers applying external stimuli are testing neither the voluntary circuit, nor Libet's hypothesis about internally triggered actions.
Discussed below (In the section "Timing intentions compared to actions") is one study that has replicated many of Libet's findings, whilst addressing some of the original criticisms. It is also worth noting that, in 2011, Itzhak Fried replicated Libet's findings at the scale of the single neuron. This was accomplished with the help of volunteer epilepsy
patients, who needed electrodes implanted deep in their brain for evaluation and treatment anyway. Now able to monitor awake and moving patients, the researchers replicated the timing anomalies that were discovered by Libet and are discussed in the following study.
Matsuhashi and Hallett's study can be summarized thus. The researchers hypothesized that, if our conscious intentions are what causes movement genesis (i.e. the start of an action), then naturally, our conscious intentions should always occur before any movement has begun. Otherwise, if we ever become aware of a movement only after it has already been started, our awareness could not have been the cause of that particular movement. Simply put, conscious intention must precede action if it is its cause.
To test this hypothesis, Matsuhashi and Hallet had volunteers perform brisk finger movements at random intervals, while not counting or planning when to make such (future) movements, but rather immediately making a movement as soon as they thought about it. An externally controlled "stop-signal" sound was played at pseudo random intervals, and the volunteers had to cancel their intent to move if they heard a signal while being aware of their own immediate intention to move. Whenever there was an action (finger movement), the authors documented (and graphed) any tones that occurred before that action. The graph of tones before actions therefore only shows tones (a) before the subject is even aware of their "movement genesis" (or else they would have stopped or "vetoed" the movement), and (b) after it is too late to veto the action. This second set of graphed tones is of little importance here.
In this work, "movement genesis" is defined as the brain process of making movement, of which physiological observations have been made (via electrodes) indicating that it may occur before conscious awareness of intent to move (see Benjamin Libet
).
By looking to see when tones started preventing actions, the researchers supposedly know the length of time (in seconds) that exists between when a subject holds a conscious intention to move and performs the action of movement. This moment of awareness (as seen in the graph below) is dubbed "T" (the mean time of conscious intention to move). It can be found by looking at the border between tones and no tones. This enables the researchers to estimate the timing of the conscious intention to move without relying on the subject's knowledge or demanding them to focus on a clock. The last step of the experiment is to compare time T for each subject with their Event-related potential
(ERP) measures (e.g. seen in this page's lead image), which reveal when their finger movement genesis first begins.
The researchers found that the time of the conscious intention to move T normally occurred too late to be the cause of movement genesis. See the example of a subject's graph below on the right. Although it is not shown on the graph, the subject's readiness potentials (ERP) tells us that his actions start at -2.8 seconds, and yet this is substantially earlier than his conscious intention to move, time "T" (-1.8 seconds). Matsuhashi and Hallet concluded that the feeling of the conscious intention to move does not cause movement genesis; both the feeling of intention and the movement itself are the result of unconscious processing.
Testing the hypothesis that 'conscious intention occurs after movement genesis has already begun' required the researchers to analyse the distribution of tones before actions. The idea is that, after time T, tones will lead to vetoing and thus a reduced representation in the data. There would also be a point of no return P where a tone was too close to the movement onset for the movement to be vetoed. In other words, the researchers were expecting to see the following on the graph: many tones (while the subjects are not yet aware of their movement genesis), followed by a drop in the number of tones during a certain period of time (while the subjects are conscious of their intentions and are stopping any movements), and finally a brief increase again in tones (when the subjects do not have the time to process the tone and prevent an action - they have past the action's "point of no return"). That is exactly what the researchers found (see the graph on the right, below).
Take one volunteer, for instance. The graph shows the times at which tones occurred when the volunteer moved. He or she showed many tones on average up until 1.8 seconds before movement onset, but a significant decrease in tones immediately after that time. Presumably this is because the subject usually became aware of his or her intention to move at about -1.8 seconds, which is then labelled point T. Since most actions are vetoed if a tone occurs after point T, there are very few tones represented during that range. Finally, there is a sudden increase in the number of tones appearing before an action at -0.1 seconds: meaning this subject has passed point P. Matsuhashi and Hallet were thus able to establish an average time T (−1.8 seconds) without subjective report. This, they compared to ERP
measurements of movement (which detected movement beginning at about −2.8 seconds on average for this participant). In the end, since T — like Libet’s original W — was often found after movement genesis had already begun, the authors concluded that the generation of awareness occurs afterwards or in parallel to action, but most importantly, it is probably not the cause.
has implications for free will and so this ability must also be explained by any theories of unconscious decision making. Philosopher Alfred Mele
also doubts the conclusions of these studies. He explains that simply because a movement may have been initiated before our "conscious self" has become aware of it does not mean our consciousness does not still get to approve, modify, and perhaps cancel (called vetoing) the action.
regions that are involved in movement planning using transcranial magnetic stimulation
in either the left or right hemisphere of the brain.
Right-handed people would normally choose to move their right hand 60% of the time, but when the right hemisphere was stimulated they would instead choose their left hand 80% of the time (recall that the right hemisphere of the brain is responsible for the left side of the body, and the left hemisphere for the right). Despite the external influence on their decision-making, the subjects continued to report that they believed their choice of hand had been made freely. In a follow-up experiment, Alvaro Pascual-Leone
and colleagues found similar results, but also noted that the transcranial magnetic stimulation must occur within 200 milliseconds, consistent with the time-course derived from the Libet experiments.
It should be noted that, despite his findings, Libet himself did not interpret his experiment as evidence of the inefficacy of conscious free will - he points out that although the tendency to press a button may be building up for 500 milliseconds, the conscious will retains a right to veto any action at the last moment. According to this model, unconscious impulses to perform a volitional act are open to suppression by the conscious efforts of the subject (sometimes referred to as "free won't"). A comparison is made with a golf
er, who may swing a club several times before striking the ball. The action simply gets a rubber stamp of approval at the last millisecond. Max Velmans
argues however that "free won't" may turn out to need as much neural preparation as "free will" (see below).
Recent research by Simone Kühn and Marcel Brass suggests that our consciousness may not be what causes some actions to be vetoed at the last moment. First of all, their experiment relies on the simple idea that we ought to know when we consciously cancel an action (i.e. we should have access to that information ). Secondly, they suggest that access to this information means humans should find it easy to tell, just after completing an action, whether it was impulsive (there being no time to decide) and when there was time to deliberate (the participant decided to allow/not to veto the action). The study found evidence that subjects could not tell this important difference. This again leaves some conceptions of free will vulnerable to the introspection illusion
. The researchers interpret their results to mean that the decision to "veto" an action is determined subconsciously, just as the initiation of the action may have been subconscious in the first place.
The primary response trials were then modified, in which 25% of the go-signals were subsequently followed by an additional signal - either a “stop” or “decide” signal. The additional signals occurred after a "signal delay" (SD), a random amount of time up to 2 seconds after the initial go-signal. They also occurred equally, each representing 12.5% of experimental cases. These additional signals were represented by the initial stimulus changing colour (e.g. to either a red or orange light). The other 75% of go-signals were not followed by an additional signal - and was therefore considered the "default" mode of the experiment. The participants' task of responding as quickly as possible to the initial signal (i.e. pressing the "go" button) remained.
Upon seeing the initial go-signal, the participant would immediately intend on pressing the "go" button. The participant was instructed to cancel their immediate intention to press the "go" button if they saw a stop signal. The participant was instructed to select randomly (at their leisure) between either pressing the "go" button, or not pressing it, if they saw a decide signal. Those trials in which the decide signal was shown after the initial go-signal ("decide trials"), for example, required that the participants prevent themselves from acting impulsively on the initial go-signal and then decide what to do. Due to the varying delays, this was sometimes impossible (e.g. some decide signals simply appeared too late in the process of them both intending to and pressing the go button for them to be obeyed).
Those trials in which the subject reacted to the go-signal impulsively without seeing a subsequent signal show a quick RT of about 600ms. Those trials in which the decide signal was shown too late, and the participant had already enacted their impulse to press the go-button (i.e. had not decided to do so), also show a quick RT of about 600ms. Those trials in which a stop signal was shown and the participant successfully responded to it, do not show a response time. Those trials in which a decide signal was shown, and the participant decided not to press the go-button, also do not show a response time. Those trials in which a decide signal was shown, and the participant had not already enacted their impulse to press the go-button, but (in which it was theorised that they) had had the opportunity to decide what to do, show a comparatively slow RT, in this case closer to 1400ms..
The participant was asked at the end of those "decide trials" in which they had actually pressed the go-button whether they had acted impulsively (without enough time to register the decide signal before enacting their intent to press the go-button in response to the initial go-signal stimulus), or had acted based upon a conscious decision made after seeing the decide signal. Based upon the response time data however, it appears there was discrepancy between when the user thought they had had the opportunity to decide (and had therefore not acted on their impulses) - in this case deciding to press the go-button, and when they thought they had acted impulsively (based upon the initial go-signal) - where the decide signal came too late to be obeyed.
Kuhn and Brass now knew what to expect: primary response trials, any failed stop trials, and the “failed-to-decide” trials were all instances where the participant obviously acted impulsively – they would show the same quick RT. In contrast, the “successful decide” trials (where the decision was a “go” and the subject moved) should show a slower RT. Presumably, if deciding whether to veto is a conscious process, volunteers should have no trouble distinguishing impulsivity from instances of true deliberate continuation of a movement. Again, this is important since decide trials require that participants rely on self knowledge. Note that stop trials cannot test self knowledge because if the subject does act, it is obvious to them that they reacted impulsively.
In decide trials the participants, it seems, were not able to reliably identify whether they had actually had time to decide – at least, not based on internal signals. The authors explain that this result is difficult to reconcile with the idea of a conscious veto, but simple to understand if the veto is considered an unconscious process. Thus it seems that the intention to move might not only arise from the subconscious, but it may only be inhibited if the subconscious says so. This conclusion could suggest that the phenomenon of "consciousness" is more of narration than direct arbitration (i.e. unconscious processing causes all thoughts, and these thoughts are again processed subconsciously).
seems to occur during action, there also seems to be processing performed after the fact - after the action is performed - to establish the full feeling of agency. Unconscious agency processing can even alter, in the moment, how we perceive the timing of sensations or actions. Kühn and Brass apply retrospective construction to explain the two peaks in "successful decide" RT's. They suggest that the late decide trials were actually deliberated, but that the impulsive early decide trials that should have been labelled "failed to decide" were mistaken during unconscious agency processing. They say that people "persist in believing that they have access to their own cognitive processes" when in fact we do a great deal of automatic unconscious processing before conscious perception occurs.
In this case, it is by the application of the forward model that one might imagine how other consciousness processes could be the result of efferent, predictive processing. If the conscious self is the efferent copy of actions and vetoes being performed, then the consciousness is a sort of narrator of what is already occurring in the body, and an incomplete narrator at that. Haggard, summarizing data taken from recent neuron recordings, says "these data give the impression that conscious intention is just a subjective corollary of an action being about to occur". Parallel processing helps explain how we might experience a sort of contra-causal free will even if it were determined.
How the brain constructs consciousness
is still a mystery, and cracking it open would have a significant bearing on the question of free will. Numerous different models have been proposed, for example, the Multiple Drafts Model
which argues that there is no central Cartesian theater
where conscious experience would be represented, but rather that consciousness is located all across the brain. This model would explain the delay between the decision and conscious realization, as experiencing everything as a continuous 'filmstrip' comes behind the actual conscious decision. In contrast, there exist models of Cartesian materialism
that have gained recognition by neuroscience, implying that there actually might be special brain areas that store the contents of consciousness; this does not, however, rule out the possibility of a conscious will. Other models such as epiphenomenalism
argue that conscious will is an illusion, and that consciousness is a by-product of physical states of the world. Work in this sector is still highly speculative, and there is no single model of consciousness which would be favored by the researchers. (See also: Philosophy of mind
.)
Although humans clearly make choices, the role of consciousness (at least, when it comes to motor movements) may need re-conceptualization. Only one thing is certain: the correlation of a conscious "intention to move" with a subsequent "action" does not guarantee causation
. Recent studies cast doubt on such a causal relation, and so more empirical data is required.
(pre-SMA) of the brain. Directly stimulating the pre-SMA causes volunteers to report a feeling of intention, and sufficient stimulation of that same area causes physical movement. This suggests that awareness of an intention to move may literally be the “sensation” of the body’s early movement, but certainly not the cause. Other studies have at least suggested that "The greater activation of the SMA, SACC, and parietal areas during and after execution of internally generated actions suggests that an important feature of internal decisions is specific neural processing taking place during and after the corresponding action. Therefore, awareness of intention timing seems to be fully established only after execution of the corresponding action, in agreement with the time course of neural activity observed here."
More recent studies using electrodes and scanning the brain more directly have yielded further insights.
Free will
"To make my own decisions whether I am successful or not due to uncontrollable forces" -Troy MorrisonA pragmatic definition of free willFree will is the ability of agents to make choices free from certain kinds of constraints. The existence of free will and its exact nature and definition have long...
, which is a philosophical
Philosophy
Philosophy is the study of general and fundamental problems, such as those connected with existence, knowledge, values, reason, mind, and language. Philosophy is distinguished from other ways of addressing such problems by its critical, generally systematic approach and its reliance on rational...
and scientific
Science
Science is a systematic enterprise that builds and organizes knowledge in the form of testable explanations and predictions about the universe...
question as to whether, and in what sense, rational agent
Rational agent
In economics, game theory, decision theory, and artificial intelligence, a rational agent is an agent which has clear preferences, models uncertainty via expected values, and always chooses to perform the action that results in the optimal outcome for itself from among all feasible actions...
s exercise control over their actions or decisions. As it has become possible to study the living brain
Brain
The brain is the center of the nervous system in all vertebrate and most invertebrate animals—only a few primitive invertebrates such as sponges, jellyfish, sea squirts and starfishes do not have one. It is located in the head, usually close to primary sensory apparatus such as vision, hearing,...
, researchers have begun to watch decision-making processes at work. While findings may carry implications for moral responsibility
Moral responsibility
Moral responsibility usually refers to the idea that a person has moral obligations in certain situations. Disobeying moral obligations, then, becomes grounds for justified punishment. Deciding what justifies punishment, if anything, is a principle concern of ethics.People who have moral...
in general, the field remains highly controversial as there is no consensus among researchers about the significance of findings, their meaning, or what conclusions may be drawn.
Overview
One significant finding of modern studies is that a person's brain seems to commit to certain decisions before the person becomes aware of having made them (see right). Researchers have found delays of about half a second. With contemporary brain scanning technology, scientists in 2008 were able to predict with 60% accuracy whether subjects would press a button with their left or right hand up to 10 seconds before the subject became aware of having made that choice. These and other findings have led some scientists, like Patrick Haggard, to reject some forms of "free will". To be clear, no single study would disprove all forms of free will. This is because the term "free willFree will
"To make my own decisions whether I am successful or not due to uncontrollable forces" -Troy MorrisonA pragmatic definition of free willFree will is the ability of agents to make choices free from certain kinds of constraints. The existence of free will and its exact nature and definition have long...
" can encapsulate different hypotheses, each of which must be considered in light of existing empirical evidence.
It is quite likely that a large range of cognitive operations are necessary to freely press a button. Research at least suggests that our conscious self does not initiate all behaviours. Instead, the conscious self is somehow alerted to behaviours that the rest of the brain and body are already planning and performing. These findings do not forbid conscious experience from playing some moderating role (although this page does discuss studies suggesting that moderating and cancelling an action is an unconscious process too). The key is that the unconscious processes play a much larger role in behaviour, because they are what first initiate actions, whereas self awareness may only recognize that an action has been prepared.
It may be possible, then, that our intuitions about the role of our conscious "intentions" have led us astray; it may be the case that we have confused correlation with causation
Correlation does not imply causation
"Correlation does not imply causation" is a phrase used in science and statistics to emphasize that correlation between two variables does not automatically imply that one causes the other "Correlation does not imply causation" (related to "ignoring a common cause" and questionable cause) is a...
by believing that conscious awareness necessarily causes the body's movement. This possibility is bolstered by findings in neurostimulation
Neurostimulation
Neurostimulation involves modulation of the nervous system and electrically activate neurons in the body. The activation of neural elements in a part of the nervous system can be effectively facilitated by stimulation. Micro-electrodes are utilized to interface with excitable tissue in order to...
, brain damage
Brain damage
"Brain damage" or "brain injury" is the destruction or degeneration of brain cells. Brain injuries occur due to a wide range of internal and external factors...
, but also research into introspection illusions
Introspection illusion
The introspection illusion is a cognitive illusion in which people wrongly think they have direct insight into the origins of their mental states, while treating others' introspections as unreliable...
. Such illusions show that humans do not have full access to various internal processes. The discovery that humans possess a determined will would have implications for moral responsibility
Moral responsibility
Moral responsibility usually refers to the idea that a person has moral obligations in certain situations. Disobeying moral obligations, then, becomes grounds for justified punishment. Deciding what justifies punishment, if anything, is a principle concern of ethics.People who have moral...
. Neuroscientist and author Sam Harris
Sam Harris (author)
Sam Harris is an American author, and neuroscientist, as well as the co-founder and current CEO of Project Reason. He received a Bachelor of Arts in philosophy from Stanford University, before receiving a Ph.D. in neuroscience from UCLA...
believes this is the case, but says we should never have expected we had libertarian free will. He argues that "Thoughts simply arise in the brain. What else could they do? The truth about us is even stranger than we may suppose: The illusion of free will is itself an illusion".
There have been a number of problems regarding studies of free will. Particularly in earlier studies, research relies too much on the introspection of the participants (other research shows that things such as introspective estimates of event timing are not accurate). The conscious decision may be made elsewhere in the brain before conscious realization, leaving room for free will even if the conscious realization comes delayed. Many brain activity measures have been insufficient and primitive as there is no good independent brain-function measure of the conscious generation of intentions, choices, or decisions. The conclusions drawn from measurements that have been made are debatable too, as they don't necessarily tell, for example, what a sudden dip in the readings are representing. In other words, the dip might have nothing to do with unconscious decision, since many other mental processes are going on while performing the task. Some of the research mentioned here has gotten more advanced, however, even recording individual neurons in live volunteers.
These studies have only just begun to shed light on the role that consciousness plays in actions and it is too early to draw very strong conclusions about many "free will"s. It is worth noting that such experiments - so far - have dealt only with free will decisions made in short time frames (seconds) and may not have direct bearing on free will decisions made ("thoughtfully") by the subject over the course of many seconds, minutes, hours or longer.
The Libet experiment
A pioneering experiment in this field was conducted by Benjamin LibetBenjamin Libet
Benjamin Libet April 12, 1916, Chicago, Illinois - July 23, 2007, Davis, California) was a pioneering scientist in the field of human consciousness. Libet was a researcher in the physiology department of the University of California, San Francisco...
in the 1980s, in which he asked each subject to choose a random moment to flick their wrist while he measured the associated activity in their brain (in particular, the build-up of electrical signal called the readiness potential). Although it was well known that the readiness potential preceded the physical action, Libet asked how the readiness potential corresponded to the felt intention to move. To determine when the subject felt the intention to move, he asked her to watch the second hand of a clock and report its position when she felt that she had felt the conscious will to move.
Libet found that the unconscious brain activity leading up to the conscious decision by the subject to flick his or her wrist began approximately half a second before the subject consciously felt that she had decided to move. Libet's findings suggest that decisions made by a subject are first being made on a subconscious level and only afterward being translated into a "conscious decision", and that the subject's belief that it occurred at the behest of her will was only due to her retrospective perspective on the event.
The interpretation of these findings has been criticized by Daniel Dennett
Daniel Dennett
Daniel Clement Dennett is an American philosopher, writer and cognitive scientist whose research centers on the philosophy of mind, philosophy of science and philosophy of biology, particularly as those fields relate to evolutionary biology and cognitive science. He is currently the Co-director of...
, who argues that people will have to shift their attention from their intention to the clock, and that this introduces temporal mismatches between the felt experience of will and the perceived position of the clock hand. Consistent with this argument, subsequent studies have shown that the exact numerical value varies depending on attention. Despite the differences in the exact numerical value, however, the main finding has held. Philosopher Alfred Mele
Alfred Mele
Alfred Remen Mele is an American philosopher. He has been the William H. and Lucyle T. Werkmeister Professor of Philosophy at Florida State University since 2000...
criticizes this design for other reasons. Having attempted the experiment himself, Mele explains that "the awareness of the intention to move" is an ambiguous feeling at best. For this reason he remained skeptical of interpreting the subjects' reported times for comparison with their 'readiness potential'.
Investigating Libet's findings
Lateralized readiness potential
In neuroscience, the lateralized readiness potential is an event-related brain potential, or increase in electrical activity at the surface of the brain, that is thought to reflect the preparation of motor activity on a certain side of the body; in other words, it is a spike in the electrical...
" (LRP), an ERP
Event-related potential
An event-related potential is any measured brain response that is directly the result of a thought or perception. More formally, it is any stereotyped electrophysiological response to an internal or external stimulus....
component which measures the difference between left and right hemisphere brain activity. Haggard and Eimer argue that the feeling of conscious will must therefore follow the decision of which hand to move, since the LRP reflects the decision to lift a particular hand.
A more direct test of the relationship between the readiness potential and the "awareness of the intention to move" was conducted by Banks and Isham (2009). In their study, participants performed a variant of the Libet's paradigm in which a delayed tone followed the button press. Subsequently, research participants reported the time of their intention to act (e.g., Libet's "W"). If W were time-locked to the readiness potential, W would remain uninfluenced by any post-action information. However, findings from this study show that W in fact shift systematically with the time of the tone presentation, implicating that W is, at least in part, retrospectively reconstructed rather than pre-determined by the readiness potential.
A study conducted by Jeff Miller and Judy Trevena (2009) suggests that the readiness potential (RP) signal in Libet's experiments doesn't represent a decision to move, but that it's merely a sign that the brain is paying attention. In this experiment the classical Libet experiment was modified by playing an audio tone indicating to volunteers to decide whether to tap a key or not. The researchers found that there was the same RP signal in both cases, regardless of whether or not volunteers actually elected to tap, which suggests that the RP signal doesn't indicate that a decision has been made.
In a second experiment, researchers asked volunteers to decide on the spot whether to use left hand or right to tap the key while monitoring their brain signals, and they found no correlation among the signals and the chosen hand. This criticism has itself been criticized by free-will researcher Patrick Haggard, who mentions literature that distinguishes two different circuits in the brain that lead to action: a "stimulus-response" circuit and a "voluntary" circuit. According to Haggard, researchers applying external stimuli are testing neither the voluntary circuit, nor Libet's hypothesis about internally triggered actions.
Discussed below (In the section "Timing intentions compared to actions") is one study that has replicated many of Libet's findings, whilst addressing some of the original criticisms. It is also worth noting that, in 2011, Itzhak Fried replicated Libet's findings at the scale of the single neuron. This was accomplished with the help of volunteer epilepsy
Epilepsy
Epilepsy is a common chronic neurological disorder characterized by seizures. These seizures are transient signs and/or symptoms of abnormal, excessive or hypersynchronous neuronal activity in the brain.About 50 million people worldwide have epilepsy, and nearly two out of every three new cases...
patients, who needed electrodes implanted deep in their brain for evaluation and treatment anyway. Now able to monitor awake and moving patients, the researchers replicated the timing anomalies that were discovered by Libet and are discussed in the following study.
Timing intentions compared to actions
A recent study by Masao Matsuhashi and Mark Hallett claims to have replicated Libet's findings without relying on subjective report or clock memorization on the part of participants. The authors believe that their method can identify the time (T) at which a subject becomes aware of her own movement. Matsuhashi and Hallet argue that this time not only varies, but often occurs after early phases of movement genesis have already begun (as measured by the readiness potential). They conclude that a person's awareness cannot be the cause of movement, and may instead only notice the movement.The experiment
To test this hypothesis, Matsuhashi and Hallet had volunteers perform brisk finger movements at random intervals, while not counting or planning when to make such (future) movements, but rather immediately making a movement as soon as they thought about it. An externally controlled "stop-signal" sound was played at pseudo random intervals, and the volunteers had to cancel their intent to move if they heard a signal while being aware of their own immediate intention to move. Whenever there was an action (finger movement), the authors documented (and graphed) any tones that occurred before that action. The graph of tones before actions therefore only shows tones (a) before the subject is even aware of their "movement genesis" (or else they would have stopped or "vetoed" the movement), and (b) after it is too late to veto the action. This second set of graphed tones is of little importance here.
In this work, "movement genesis" is defined as the brain process of making movement, of which physiological observations have been made (via electrodes) indicating that it may occur before conscious awareness of intent to move (see Benjamin Libet
Benjamin Libet
Benjamin Libet April 12, 1916, Chicago, Illinois - July 23, 2007, Davis, California) was a pioneering scientist in the field of human consciousness. Libet was a researcher in the physiology department of the University of California, San Francisco...
).
By looking to see when tones started preventing actions, the researchers supposedly know the length of time (in seconds) that exists between when a subject holds a conscious intention to move and performs the action of movement. This moment of awareness (as seen in the graph below) is dubbed "T" (the mean time of conscious intention to move). It can be found by looking at the border between tones and no tones. This enables the researchers to estimate the timing of the conscious intention to move without relying on the subject's knowledge or demanding them to focus on a clock. The last step of the experiment is to compare time T for each subject with their Event-related potential
Event-related potential
An event-related potential is any measured brain response that is directly the result of a thought or perception. More formally, it is any stereotyped electrophysiological response to an internal or external stimulus....
(ERP) measures (e.g. seen in this page's lead image), which reveal when their finger movement genesis first begins.
The researchers found that the time of the conscious intention to move T normally occurred too late to be the cause of movement genesis. See the example of a subject's graph below on the right. Although it is not shown on the graph, the subject's readiness potentials (ERP) tells us that his actions start at -2.8 seconds, and yet this is substantially earlier than his conscious intention to move, time "T" (-1.8 seconds). Matsuhashi and Hallet concluded that the feeling of the conscious intention to move does not cause movement genesis; both the feeling of intention and the movement itself are the result of unconscious processing.
The rationale
This study is similar to Libet's in some ways: volunteers were again asked to perform finger extensions in short, self-paced intervals. In this version of the experiment, researchers introduced randomly timed “stop tones” during the self paced movements. If participants were not conscious of any intention to move, they simply ignored the tone. On the other hand, if they were aware of their intention to move at the time of the tone, they had to try to veto the action, then relax for a bit before continuing self-paced movements. This experimental design allowed Matsuhashi and Hallet to see when, once the subject moved their finger, any tones occurred. The goal was to identify their own equivalent of Libet’s W, their own estimation of the timing of the conscious intention to move, which they would call “T”.Testing the hypothesis that 'conscious intention occurs after movement genesis has already begun' required the researchers to analyse the distribution of tones before actions. The idea is that, after time T, tones will lead to vetoing and thus a reduced representation in the data. There would also be a point of no return P where a tone was too close to the movement onset for the movement to be vetoed. In other words, the researchers were expecting to see the following on the graph: many tones (while the subjects are not yet aware of their movement genesis), followed by a drop in the number of tones during a certain period of time (while the subjects are conscious of their intentions and are stopping any movements), and finally a brief increase again in tones (when the subjects do not have the time to process the tone and prevent an action - they have past the action's "point of no return"). That is exactly what the researchers found (see the graph on the right, below).
Take one volunteer, for instance. The graph shows the times at which tones occurred when the volunteer moved. He or she showed many tones on average up until 1.8 seconds before movement onset, but a significant decrease in tones immediately after that time. Presumably this is because the subject usually became aware of his or her intention to move at about -1.8 seconds, which is then labelled point T. Since most actions are vetoed if a tone occurs after point T, there are very few tones represented during that range. Finally, there is a sudden increase in the number of tones appearing before an action at -0.1 seconds: meaning this subject has passed point P. Matsuhashi and Hallet were thus able to establish an average time T (−1.8 seconds) without subjective report. This, they compared to ERP
Event-related potential
An event-related potential is any measured brain response that is directly the result of a thought or perception. More formally, it is any stereotyped electrophysiological response to an internal or external stimulus....
measurements of movement (which detected movement beginning at about −2.8 seconds on average for this participant). In the end, since T — like Libet’s original W — was often found after movement genesis had already begun, the authors concluded that the generation of awareness occurs afterwards or in parallel to action, but most importantly, it is probably not the cause.
Criticisms
Haggard describes other studies at the neuronal levels as providing "a reassuring confirmation of previous studies that recorded neural populations" such as the one just described. Note that these results were gathered using finger movements, and may not necessarily generalize to other actions such as thinking, or even other motor actions in different situations. Indeed, the human act of planningPlanning
Planning in organizations and public policy is both the organizational process of creating and maintaining a plan; and the psychological process of thinking about the activities required to create a desired goal on some scale. As such, it is a fundamental property of intelligent behavior...
has implications for free will and so this ability must also be explained by any theories of unconscious decision making. Philosopher Alfred Mele
Alfred Mele
Alfred Remen Mele is an American philosopher. He has been the William H. and Lucyle T. Werkmeister Professor of Philosophy at Florida State University since 2000...
also doubts the conclusions of these studies. He explains that simply because a movement may have been initiated before our "conscious self" has become aware of it does not mean our consciousness does not still get to approve, modify, and perhaps cancel (called vetoing) the action.
Manipulating the unconscious
Related experiments showed that neurostimulation could affect which hands people move, even though the experience of free will was intact. Ammon and Gandevia found that it was possible to influence which hand people move by stimulating frontalFrontal lobe
The frontal lobe is an area in the brain of humans and other mammals, located at the front of each cerebral hemisphere and positioned anterior to the parietal lobe and superior and anterior to the temporal lobes...
regions that are involved in movement planning using transcranial magnetic stimulation
Transcranial magnetic stimulation
Transcranial magnetic stimulation is a noninvasive method to cause depolarization or hyperpolarization in the neurons of the brain...
in either the left or right hemisphere of the brain.
Alvaro Pascual-Leone
Alvaro Pascual-Leone is a Professor of Neurology at Harvard Medical School, with which he has been affiliated since 1997. He is the Director of the Berenson-Allen Center for Noninvasive Brain Stimulation and Program Director of the Harvard-Thorndike Clinical Research Center of the Beth Israel...
and colleagues found similar results, but also noted that the transcranial magnetic stimulation must occur within 200 milliseconds, consistent with the time-course derived from the Libet experiments.
It should be noted that, despite his findings, Libet himself did not interpret his experiment as evidence of the inefficacy of conscious free will - he points out that although the tendency to press a button may be building up for 500 milliseconds, the conscious will retains a right to veto any action at the last moment. According to this model, unconscious impulses to perform a volitional act are open to suppression by the conscious efforts of the subject (sometimes referred to as "free won't"). A comparison is made with a golf
Golf
Golf is a precision club and ball sport, in which competing players use many types of clubs to hit balls into a series of holes on a golf course using the fewest number of strokes....
er, who may swing a club several times before striking the ball. The action simply gets a rubber stamp of approval at the last millisecond. Max Velmans
Max Velmans
Max Velmans is a Professor of Psychology at Goldsmiths, University of London. From 2003 to 2006 he was Chair of the Consciousness and Experiential Psychology Section of the British Psychological Society and has been involved in many other initiatives to encourage studies in consciousness...
argues however that "free won't" may turn out to need as much neural preparation as "free will" (see below).
Unconsciously cancelling actions
The possibility that human "free won't" is also the prerogative of the subconscious is being explored.Retrospective judgement of free choice
Introspection illusion
The introspection illusion is a cognitive illusion in which people wrongly think they have direct insight into the origins of their mental states, while treating others' introspections as unreliable...
. The researchers interpret their results to mean that the decision to "veto" an action is determined subconsciously, just as the initiation of the action may have been subconscious in the first place.
The experiment
The experiment involved asking volunteers to respond to a go-signal by pressing an electronic "go" button as quickly as possible. In this experiment the go-signal was represented as a visual stimulus shown on a monitor (e.g. a green light as shown on the picture). The participants' reaction times (RT) were gathered at this stage, in what was described as the "primary response trials".The primary response trials were then modified, in which 25% of the go-signals were subsequently followed by an additional signal - either a “stop” or “decide” signal. The additional signals occurred after a "signal delay" (SD), a random amount of time up to 2 seconds after the initial go-signal. They also occurred equally, each representing 12.5% of experimental cases. These additional signals were represented by the initial stimulus changing colour (e.g. to either a red or orange light). The other 75% of go-signals were not followed by an additional signal - and was therefore considered the "default" mode of the experiment. The participants' task of responding as quickly as possible to the initial signal (i.e. pressing the "go" button) remained.
Upon seeing the initial go-signal, the participant would immediately intend on pressing the "go" button. The participant was instructed to cancel their immediate intention to press the "go" button if they saw a stop signal. The participant was instructed to select randomly (at their leisure) between either pressing the "go" button, or not pressing it, if they saw a decide signal. Those trials in which the decide signal was shown after the initial go-signal ("decide trials"), for example, required that the participants prevent themselves from acting impulsively on the initial go-signal and then decide what to do. Due to the varying delays, this was sometimes impossible (e.g. some decide signals simply appeared too late in the process of them both intending to and pressing the go button for them to be obeyed).
Those trials in which the subject reacted to the go-signal impulsively without seeing a subsequent signal show a quick RT of about 600ms. Those trials in which the decide signal was shown too late, and the participant had already enacted their impulse to press the go-button (i.e. had not decided to do so), also show a quick RT of about 600ms. Those trials in which a stop signal was shown and the participant successfully responded to it, do not show a response time. Those trials in which a decide signal was shown, and the participant decided not to press the go-button, also do not show a response time. Those trials in which a decide signal was shown, and the participant had not already enacted their impulse to press the go-button, but (in which it was theorised that they) had had the opportunity to decide what to do, show a comparatively slow RT, in this case closer to 1400ms..
The participant was asked at the end of those "decide trials" in which they had actually pressed the go-button whether they had acted impulsively (without enough time to register the decide signal before enacting their intent to press the go-button in response to the initial go-signal stimulus), or had acted based upon a conscious decision made after seeing the decide signal. Based upon the response time data however, it appears there was discrepancy between when the user thought they had had the opportunity to decide (and had therefore not acted on their impulses) - in this case deciding to press the go-button, and when they thought they had acted impulsively (based upon the initial go-signal) - where the decide signal came too late to be obeyed.
The rationale
Kuhn and Brass wanted to test participant self-knowledge. The first step was that after every decide trial, participants were next asked whether they had actually had time to decide. Specifically, the volunteers were asked to label each decide trial as either failed-to-decide (the action was the result of acting impulsively on the initial go-signal) or successful decide (the result of a deliberated decision). See the diagram on the right for this decide trial split: failed-to-decide and successful decide; the next split in this diagram (participant correct or incorrect) will be explained at the end of this experiment. Note also that the researchers sorted the participants’ successful decide trials into “decide go” and “decide nogo”, but were not concerned with the nogo trials since they did not yield any RT data (and are not featured anywhere in the diagram on the right). Note that successful stop trials did not yield RT data either.Kuhn and Brass now knew what to expect: primary response trials, any failed stop trials, and the “failed-to-decide” trials were all instances where the participant obviously acted impulsively – they would show the same quick RT. In contrast, the “successful decide” trials (where the decision was a “go” and the subject moved) should show a slower RT. Presumably, if deciding whether to veto is a conscious process, volunteers should have no trouble distinguishing impulsivity from instances of true deliberate continuation of a movement. Again, this is important since decide trials require that participants rely on self knowledge. Note that stop trials cannot test self knowledge because if the subject does act, it is obvious to them that they reacted impulsively.
Results and implications
Unsurprisingly, the recorded RTs for the primary response trials, failed stop trials, and “failed-to-decide” trials all showed similar RTs: 600ms seems to indicate an impulsive action made without time to truly deliberate. What the two researchers found next was not as easy to explain: while some “successful decide” trials did show the tell-tale slow RT of deliberation (averaging around 1400ms), participants had also labelled many impulsive actions as “successful decide”. This result is startling because participants should have had no trouble identifying which actions were the results of a conscious “I will not veto”, and which actions were un-deliberated, impulsive reactions to the initial go-signal. As the authors explain:In decide trials the participants, it seems, were not able to reliably identify whether they had actually had time to decide – at least, not based on internal signals. The authors explain that this result is difficult to reconcile with the idea of a conscious veto, but simple to understand if the veto is considered an unconscious process. Thus it seems that the intention to move might not only arise from the subconscious, but it may only be inhibited if the subconscious says so. This conclusion could suggest that the phenomenon of "consciousness" is more of narration than direct arbitration (i.e. unconscious processing causes all thoughts, and these thoughts are again processed subconsciously).
Criticisms
In this recent experiment, subjects could not reliably distinguish between "producing an action without stopping and stopping an action before voluntarily resuming". The conclusions that are drawn from this information, however - such as the assumption that only the faster decisions can be unconscious and that because the conscious ones have the same timing they must also be unconscious - have yet to be debated properly or even replicated. It is also worth noting that Libet consistently referred to a veto of an action that was initiated endogenously. That is, a veto that occurs in the absence of external cues, instead relying on only internal cues (if any at all). This veto may be a different type of veto than the one explored by Kühn and Brass using their decide signal.Retrospective construction
The idea behind retrospective construction is that, while part of the "yes, I did it" feeling of agencyAgency (sociology)
In the social sciences, agency refers to the capacity of individuals to act independently and to make their own free choices. By contrast, "Structure" refers to the factors of influence that determine or limit an agent and his or her decisions...
seems to occur during action, there also seems to be processing performed after the fact - after the action is performed - to establish the full feeling of agency. Unconscious agency processing can even alter, in the moment, how we perceive the timing of sensations or actions. Kühn and Brass apply retrospective construction to explain the two peaks in "successful decide" RT's. They suggest that the late decide trials were actually deliberated, but that the impulsive early decide trials that should have been labelled "failed to decide" were mistaken during unconscious agency processing. They say that people "persist in believing that they have access to their own cognitive processes" when in fact we do a great deal of automatic unconscious processing before conscious perception occurs.
Related models
The idea that intention co-occurs (rather than causes) movement is reminiscent of "forward models of motor control" (or FMMC, which have been used to try to explain inner speech). FMMCs describe parallel circuits: movement is processed in parallel with other predictions of movement; if the movement matches the prediction - the feeling of agency occurs. FMMCs have been applied in other related experiments. Metcalfe and her colleagues used an FMMC to explain how volunteers determine whether they are in control of a computer game task. On the other hand, they acknowledge other factors too. The authors attribute feelings of agency to desirability of the results (see self serving biases) and top-down processing (reasoning and inferences about the situation).In this case, it is by the application of the forward model that one might imagine how other consciousness processes could be the result of efferent, predictive processing. If the conscious self is the efferent copy of actions and vetoes being performed, then the consciousness is a sort of narrator of what is already occurring in the body, and an incomplete narrator at that. Haggard, summarizing data taken from recent neuron recordings, says "these data give the impression that conscious intention is just a subjective corollary of an action being about to occur". Parallel processing helps explain how we might experience a sort of contra-causal free will even if it were determined.
How the brain constructs consciousness
Consciousness
Consciousness is a term that refers to the relationship between the mind and the world with which it interacts. It has been defined as: subjectivity, awareness, the ability to experience or to feel, wakefulness, having a sense of selfhood, and the executive control system of the mind...
is still a mystery, and cracking it open would have a significant bearing on the question of free will. Numerous different models have been proposed, for example, the Multiple Drafts Model
Multiple Drafts Model
Daniel Dennett's multiple drafts model of consciousness is a physicalist theory of consciousness based upon cognitivism, which views the mind in terms of information processing. The theory is described in depth in his book, Consciousness Explained, published in 1991...
which argues that there is no central Cartesian theater
Cartesian theater
The Cartesian theater is a derisive term coined by philosopher Daniel Dennett to pointedly refer to a defining aspect of what he calls Cartesian materialism, which he considers to be the often unacknowledged remnants of Cartesian dualism in modern materialistic theories of the mind.Descartes...
where conscious experience would be represented, but rather that consciousness is located all across the brain. This model would explain the delay between the decision and conscious realization, as experiencing everything as a continuous 'filmstrip' comes behind the actual conscious decision. In contrast, there exist models of Cartesian materialism
Cartesian materialism
In philosophy of mind, Cartesian materialism is the idea that at some place in the brain, there is some set of information that directly corresponds to our conscious experience...
that have gained recognition by neuroscience, implying that there actually might be special brain areas that store the contents of consciousness; this does not, however, rule out the possibility of a conscious will. Other models such as epiphenomenalism
Epiphenomenalism
In philosophy of mind, epiphenomenalism, also known as Type-E Dualism, is a view that "mental" states do not have any influence on "physical" states.-Background:...
argue that conscious will is an illusion, and that consciousness is a by-product of physical states of the world. Work in this sector is still highly speculative, and there is no single model of consciousness which would be favored by the researchers. (See also: Philosophy of mind
Philosophy of mind
Philosophy of mind is a branch of philosophy that studies the nature of the mind, mental events, mental functions, mental properties, consciousness and their relationship to the physical body, particularly the brain. The mind-body problem, i.e...
.)
Although humans clearly make choices, the role of consciousness (at least, when it comes to motor movements) may need re-conceptualization. Only one thing is certain: the correlation of a conscious "intention to move" with a subsequent "action" does not guarantee causation
Correlation does not imply causation
"Correlation does not imply causation" is a phrase used in science and statistics to emphasize that correlation between two variables does not automatically imply that one causes the other "Correlation does not imply causation" (related to "ignoring a common cause" and questionable cause) is a...
. Recent studies cast doubt on such a causal relation, and so more empirical data is required.
The pre-SMA and the intention to move
Scanning the brain in action, the readiness potential that indicates the beginning of movement genesis is recorded by an EEG applied to the pre-supplementary motor areaSupplementary motor area
The supplementary motor area is a part of the sensorimotor cerebral cortex . It was included, on purely cytoarchitectonic arguments, in area 6 of Brodmann and the Vogts...
(pre-SMA) of the brain. Directly stimulating the pre-SMA causes volunteers to report a feeling of intention, and sufficient stimulation of that same area causes physical movement. This suggests that awareness of an intention to move may literally be the “sensation” of the body’s early movement, but certainly not the cause. Other studies have at least suggested that "The greater activation of the SMA, SACC, and parietal areas during and after execution of internally generated actions suggests that an important feature of internal decisions is specific neural processing taking place during and after the corresponding action. Therefore, awareness of intention timing seems to be fully established only after execution of the corresponding action, in agreement with the time course of neural activity observed here."
More recent studies using electrodes and scanning the brain more directly have yielded further insights.