Abstract
Wegner’s theory of conscious will applies Michotte’s causal perception model to people’s consciousness of will. The model proposes that people’s perception of conscious will reflects causal inferences about the temporal priority of the intention, the consistency of the intention with the action, and the exclusivity of their intention as the cause. This model has generated much discussion, but few commentators have examined the attribution model underpinning the theory. This analysis examines this theory of conscious will and relevant research, and finds that the research fails to provide strong support for the theory. Several conceptual propositions assumed by the theory also lack support. Research supports alternative models of intentionality that construe conscious cognitions as causes as well as effects.
Keywords
Do people have free will and if so what processes does it involve? If people don’t have free will, then what accounts for people’s conscious impression of will and self-efficacy? These are old questions in philosophy and psychology, but they are being addressed with new vigor, new paradigms, and new methods. Wegner (e.g., 2002, 2005, 2008) recently proposed a novel account of conscious will, arguing that this cognition is illusory. The model claims that people’s consciousness of will reflects the same causal relations as the perception of physical events, rather than access to causes of their behavior.
Wegner’s model has generated much discussion, including 23 commentaries in a recent issue of Behavioral and Brain Sciences (Bayne, 2006; Malle, 2006; Nahmias, 2002, 2005). However, despite the fact that Michotte’s model of causal perception forms the core of Wegner’s theory (Wegner, 2004, p. 685), few commentaries have systematically examined this attributional underpinning. This paper addresses this important gap in the discussion and examines the attribution model and research underpinning Wegner’s theory. Comment is also made on the metatheoretical propositions in Wegner’s analysis. I begin by placing Wegner’s argument in historical context.
The replacement of teleological concepts in science and social psychology
Many scientific analyses have generated causal explanations for events that previously were explained by teleological concepts such as a God (Rychlak, 1968). Thus events like thunder and illness that were interpreted as acts of gods are attributed to natural causes. Naturalistic explanations extended from the physical sciences to disciplines involving human action: biology and psychology (Rychlak, 1968). Skinner (1971) claimed that human behavior reflects the same causal principles as other species, and replaced concepts like freedom and dignity with causal analyses.
Neuroscience research on brain injury and dementia has shown that changes to a person’s brain affect every aspect of their personality, and has increasingly mapped psychological processes onto brain functions (e.g., Delacour, 1995). With intentional actions, Libet (1985) showed that when people move their finger, the readiness potentials in their brain precede conscious awareness of movement, suggesting that the action’s cause is unconscious. This extension of causal concepts to intentional action has not lacked dissenters. Even among researchers who accept naturalistic explanations, Skinner’s and Libet’s claims are challenged (e.g., Mele, 2009; Moore, Lagnado, Deal, & Haggard, 2009; Passingham & Lau, 2006).
In social psychology, researchers have shown that people often misunderstand the causes of their behavior. In their very influential review (Baumeister, 2008), Nisbett and Wilson (1977) famously concluded that people misattribute their own behavior and their self-attributions do not differ from observers’ attributions: “If the reports of subjects do not differ from the reports of observers, then it is unnecessary to assume that the former are drawing on a ‘fount of privileged knowledge’” (p. 248).
Authors who cite this review as decisive overlook objections to Nisbett and Wilson’s conclusions (e.g., Ericsson & Simon, 1980; Malle, 2006; McClure, 1983, 1991; Sabini & Silver, 1981; White 1980). For example, Nisbett and Wilson focused on retrospective reports about cognitions such as emotions, whereas actors’ concurrent reports of intentions and their prospective predictions of their actions show greater accuracy (Ericsson & Simon, 1980; Haggard & Clark, 2003; Malle, 2006; McClure, 1983, 1991; White, 1980). The case against introspective access is less clear-cut than many assume.
This observation applies to demonstrations that people are influenced by messages outside of awareness. For example, Bargh, Chen, and Burrows (1996) showed that participants primed with words associated with elderly persons subsequently walked more slowly than a control group, but were unaware of this effect. Similar findings have been obtained with other actions and cognitions, showing that unconscious influences on behavior are extensive. However, as with Nisbett and Wilson (1977), most studies do not examine the possible effect of intentions on behavior.
In contrast, Aarts and colleagues’ research applies priming to intentions and agency and is thus relevant to the analysis of will. For example, when Aarts, Custers, and Marien (2008) triggered unconscious motivation by subliminally priming the concept of physical exertion, people spent more effort on a physical measure (handgrip force) than in a non-primed condition. Aarts, Custers, and Wegner (2005) linked this priming paradigm to Michotte’s (1963) model. However, apart from presenting primes prior to actions, this research does not directly depend on Michotte’s (1963) three parameters of perception causation, unlike Wegner’s theory, which is analyzed here.
Wegner’s theory: The illusion of conscious will
Whereas most research on introspection examines people’s awareness of unconscious causes of their behavior, Wegner (e.g., 2002, 2005, 2008) extends the analysis to consciousness and will. Wegner’s theory that people’s consciousness of will is illusory has two features that comprise a novel contribution to the analysis of will. First, Wegner (2002) integrates numerous social phenomena where people misattribute their agency as a cause of events. Second, he performed experiments showing that people feel will or control over events with other causes. As Wegner (2004, p. 685) observed, the core idea in his analysis is that the experience of will reflects the properties of causal perception that Michotte (1963) applied to physical causes. Michotte’s model develops Hume’s (1777/1900) theory that people’s self-perceptions reflect sense impressions, as with external events.
Other researchers who examined the causal relations underpinning attributions of intention argue that these cognitions involve different processes than the perception of physical causes. Heider (1958) proposed that people’s perception of intentional actions is characterized by equifinality and “local causation,” whereby actions require no external cause. Heider’s model implies that perceptions of intentionality differ from perceptions of physical causes. Michotte (1963, pp. 212–214) also claimed that people’s perception of their voluntary actions uses different processes than perceptions of physical causes.
Other research has shown that certain movements of physical objects generate perceptions of intentional (or social) causality rather than physical causality (Heider & Simmel, 1944; Schlottmann, Ray, Mitchell, & Demetriou, 2006). For example, when a first object moves toward a second object and the second object moves before a first object has touched it, perceivers see the movement of the second object as reacting or “running away,” and may see the first object as “chasing” the second.
In contrast with this tradition, Wegner (e.g., 2002) applies Michotte’s (1963) model of the causal perception of physical events to the consciousness of will. Michotte claimed that three features of physical events determine whether people see one event as causing another. First, to be seen as a cause, an event must show temporal priority, as where a billiard ball strikes a second ball before the second ball moves. Second, the event must show consistency with the outcome, as where the second billiard ball moves in the same direction as the first. Third, the event must be the exclusive cause of the outcome, and multiple prior events render the cause of the movement uncertain.
Wegner (e.g., 2002) claims that Michotte’s three principles of physical causation determine people’s perception of their own actions. Thus the priority principle implies that when a thought immediately precedes an action, people tend to see that thought as causing the action. The consistency relation implies that when a thought is compatible with the action, people tend to see it as the cause. The exclusivity parameter implies that when a thought is the only apparent cause of an action, people tend to see the event as a cause of our action, whereas when there are other plausible causes, people discount the thought as an explanation. Dramatic examples of alternative explanations of actions to will are evident in hypnosis, where agency appears to transfer to the hypnotist, and Milgram’s (1963) experiments on obedience, where agency is subjugated to the experimenter’s orders. Wegner links this exclusivity concept to Kelley’s (1972) discounting principle and claims support from McClure’s (1998) analysis of discounting. Wegner’s model implies that actions result from a single cause that is negatively related to other causes.
Wegner (e.g., 2002) supported this theory with ingenious experiments that manipulated these causal principles so people think they cause events that are actually controlled by the experimenter. Wegner, Sparrow, and Winerman (2004) demonstrated the priority and consistency principles by placing participants in two roles. The participant was dressed in a smock facing a mirror, while the “hand helper” stood behind the participant and was hidden by a screen. The hand helpers put their arms through the smock so they looked like the participants’ arms, whereas the participants were told to keep their arms at their side. Instructions were given through headphones to make movements, such as snap their fingers. In one condition, only hand helpers heard the instructions, whereas in a second condition, the participant also heard the instructions. Thus participants heard a prior “thought” consistent with movements that looked like their own actions. Experiments varied the time relation between the instructions and actions.
In these studies, participants reported more control and conscious will over the arm movement in conditions where they heard the instructions. Wegner et al. (2004) claimed that this study demonstrates both the consistency principle, in that consistent “thoughts” (the instructions) and actions increased control judgments, and the priority principle, because the effect occurred only when instructions immediately preceded the movements, and not when instructions occurred long before the action or after the action. Wegner (2008) drew an analogy to times when someone telephones just when we are thinking about them—we feel like we conjured them up to ring.
Other experiments examined whether people attribute their actions to their intentions when other causes are absent (exclusivity). In Wegner and Wheatley’s (1999) experiments, participants and an experimental confederate both controlled a mouse-driven computer pointer that they could stop on target pictures on a screen. They received instructions to stop the pointer at various time intervals before and after the pointer stopped. When the pointer stopped on the target pictures immediately following these instructions, participants inferred that they intended the pointer to stop, even though the confederate had stopped the pointer.
Wegner (e.g., 2002) argues that these causal relations explain natural situations where people mistakenly infer that they caused an outcome, such as movements on an Ouija board. Wegner also invokes the causal inferences exploited in magic tricks like pulling a rabbit out of a hat. As Kelley (1980) observed, magicians conceal the real causal sequence behind an outcome and create an illusory alternative sequence. Wegner claims that the impression of will is no more evidence for causation than the impression that the rabbit comes out of a hat.
Evaluation of Wegner’s causal perception theory
Previous commentaries
This research by Wegner and his colleagues shows that people can make errors in their inferences about their will and their control over events. Several researchers have claimed that it does not follow that the impression of will is illusory (e.g., Ainslie, 2004; Bayne, 2006). They observe that when people experience other perceptual illusions, researchers do not conclude that perception is necessarily illusory or inaccurate. They conclude that Wegner’s conclusions from misperceptions of will are unjustified.
Few researchers however, have challenged Wegner’s application of Michotte’s (1963) model to people’s consciousness of will. One exception is Schultz, Sebanz, and Frith (2004), who evaluated the theory in relation to patients with schizophrenia who experienced delusions (see also Bayne, 2006). First, they claim that these patients’ view of their actions is consistent with the three Michotte parameters, because they have prior intentions to perform actions consistent with their intentions, and there is no ambiguity about who is acting. Yet the patients think their actions are controlled by an alien force. This is a legitimate observation, but Wegner could claim that these actions fail to show exclusivity, because they attributed actions to the alien.
Second, Schultz et al. claim that Wegner’s examples differ from everyday life in that they show a lack of exclusivity, and “people are tricked into attributing to themselves an intention that they never had” (2004, p. 675). This point is valid, but may not undermine Wegner’s argument, as it is consistent with his claim that causes are discounted when other factors influence the outcome. Many real-world situations similarly lack exclusive causes. Third, Schultz et al. note that Wegner’s experiments involve intentions in action rather than premeditated actions, whereas in everyday life, many actions follow from prior intentions and deliberation. This point poses a serious challenge to Wegner’s theory which is examined more closely here.
The effects of priority and consistency information
There are several problems in Wegner’s application of Michotte’s (1963) three parameters of causal perception to will and intentional action. The first relation, the priority principle, holds that people perceive causality when a thought immediately precedes an action. Michotte showed that when there is an interval between a physical event and an outcome, people do not perceive the prior event to be the cause. Wegner (2002) reiterates this view: “To be perceived as truly a cause, an event can’t start too soon or too late; it has to occur just before the effect” (p. 71). Wegner applies this feature to will: “It is clear that there is only a fairly small window prior to an action in which relevant thoughts must appear if the action is to be felt as willed” (Wegner, 2002, p. 72). And again: “Thought that occurs too far in advance of an action is not likely to be seen as the cause of it” (p. 71).
Wegner presents examples that support this argument, where a thought immediately precedes an action (e.g., Wegner & Wheatley, 1999). The challenge for Wegner’s claim is that, unlike the physical events examined in the Michotte paradigm, intentions may exist long before the action occurs (Nahmias, 2002, 2005). A person may plan a sabbatical trip or a murder long before the action. At another point, Wegner (2002) acknowledged that premeditation of actions occurs, but argued that this planning only produces the experience of will if the plan reappears in mind immediately before the action: “In the absence of thought about the action just prior to the performance, even the most distant foresight would merely be premature and would do little to promote the feeling that one had willed the action” (p. 73). Wegner gives no evidence to underpin this claim, or explain why foresight is “premature.”
Wegner (2002) reinforced this argument with Gollwitzer’s claim that when actions are planned in advance to correspond with a triggering event (“I’ll go when the light turns green”), then at the appropriate time the actions occur automatically without conscious thought. It is debatable whether people consciously plan in advance to go when lights turn green. But setting aside this less than convincing example, there is no evidence that planned actions require triggering events (Nahmias, 2005). Research on attributions about causal chains shows that people see distal intentions as good explanations of events and as better explanations than physical causes leading to similar outcomes (Hilton, McClure, & Sutton, 2010; Lagnado & Channon, 2008; McClure, Hilton, & Sutton, 2007). In addition, when planned actions approach realization, people shift from broad goals to specific actions and implementations (Trope & Liberman, 2003; Vallacher & Wegner, 1987). However, there is no evidence that people think these actions are not intended. This issue suggests that perceptions of will involve different processes than inferences about physical causes.
In response to Schultz et al.’s (2004) arguments about premeditated action, Wegner (2004) modified his theory and claimed “the timing of the experience of will with regard to actions … is not particularly telling regarding the main thesis” of his theory and “just because an experience of conscious will happened well in advance of brain events leading to an action … does not mean that the experience is any more direct an indication of the causal process whereby the action was produced” (p. 684). Wegner is correct on this point, but this revision of the theory to allow a delay between intention and action conflicts with Wegner’s previous claim that events must immediately precede an outcome to be seen as a cause (e.g., Wegner, 2002).
A related issue that is not addressed in Wegner’s theory is that both physical and intentional events involve causal chains, where an initial cause may occur well before an outcome and several other causes may intervene between the initial cause and the outcome. People’s judgments reflect these causal chains and often prefer distal causes over proximal causes (Lagnado & Channon, 2008; McClure et al., 2007; O’Doherty, Navarro, & Crabb, 2009; Sternberg, 2004). Michotte’s model only deals with people’s perception of causation with two consecutive events.
Questions can also be raised concerning Wegner’s application of Michotte’s (1963) consistency principle, where he claims that people infer their action is willed when their thought is consistent with the subsequent action. Difficulties with this claim have been noted in its previous incarnations (e.g., Skinner, 1957); specifically, intentions may lead to unintended consequences and an intended outcome may result from events other than the relevant action (Giddens, 1976). There is no evidence that in either of these cases, people infer that the intention was absent. Situations where intended actions produce unintended consequences do not threaten people’s sense of will. Indeed, people take credit for unintended positive consequences of their actions (McMartin & Shaw, 1977).
Wegner (2008) recently related the consistency parameter to the evolutionary value of the consciousness of will, suggesting that this cognition was adaptive because it enabled people to signal intentions: “Communications of intention serve the purpose of making many potentially costly social actions unnecessary because the statement itself causes preventive responses” (p. 240); for example, where telling people that you intend to hit them leads them to preempt the action. In this case, stating the intention serves to prevent the action that would fulfill the intention, so it is inconsistent with the action that occurs. This idea may be plausible as a hypothesis about humans’ past, but it contradicts the consistency principle, whereby people infer intentions when a thought is consistent with a subsequent action.
The experiments by Wegner et al. (2004) showing that people’s perceptions of will are affected by the priority and consistency of causes are ingenious and interesting. The experiments showed that people in the participant role perceived some control over the hand-helpers’ movements of their arms through the smock, even though they lacked this control. Although participants’ ratings of their control over the hand helpers’ movements and willing the hands to move were higher with instructions than with no instructions, the rating in the instruction condition was still modest (a mean of 3.00 on a 7-point scale in Expt. 1, and 2.46 in Expt. 2). This paradigm shows that people may perceive some control over movements that they do not control, but it does not refute the notion that people generate intentions and perform actions that fulfill those intentions. These experiments involve small though significant changes in “feeling of control” and such. A more compelling demonstration would be if participants found themselves unable to interfere with the movements of the hand helpers (Nahmias, 2005).
A related point is that in these studies, participants were instructed to keep their arms still, and as they were in a subject role and had to perform this role cooperatively to get their course credits, they apparently did so. But it is possible that they might not obediently perform this action and instead exercise their volition and do something else. Alternatively, they could be instructed to not do what they hear on the earphones, and when they hear the instruction to click fingers they should wave their arms while the hand helpers click their fingers. If participants were then asked which of these two actions was theirs or which action reflects their will, it is unlikely they would claim authorship over the hand helpers’ actions. So this interesting experiment does not provide compelling evidence that perceptions of will are based primarily on Michotte’s principles of priority and consistency.
This point relates to Wegner’s (e.g., 2002) claim that people’s introspective illusions about their will are analogous to the extraspective illusions created by a magician, such as the perception that a rabbit comes out of a hat. Wegner’s model portrays humans as analogous to a magician’s audience. However, humans are not only similar to audiences who are tricked by magicians, they also function like magicians who trick people. People are capable of misleading others about their intentions and the real causal sequence behind their actions (Kelley, 1980). This process is central to military deception (Latimer, 2001). Psychology researchers often portray citizens as similar only to their experimental participants who are following instructions as “subjects” and do not portray them as similar to the experimenters themselves who are designing experiments and manipulating events (McClure, 1985, 1991). Danziger (1990) noted that participants often see it as their role to provide the experimenters with a desired performance, rather than act as human agents (p. 62). A theory of will needs to encompass people’s capacities in both roles: magician and audience, experimenter and participant, people who obey instructions and people who do not. Research that only understands people in the role of audiences and obedient subjects is likely to underestimate people’s capacity for intentional action.
The effects of exclusivity information
Serious questions also pertain to Wegner’s application of Michotte’s exclusivity principle, which proposes that people see their actions as caused by their intentions when other causes are absent. Researchers have challenged the implications of Wegner and Wheatley’s (1999) experiments, where participants and an experimental confederate received instructions when to stop a computer pointer at various time intervals. These experiments demonstrate participants’ errors and confabulations, but they do not simulate a situation where a person generates intentions to perform an action (Bayne, 2006; Malle, 2006).
With regard to the exclusivity parameter, Wegner (2008) recently suggested that the illusion of agency swamping causal logic is best illustrated in Woolfork, Doris, and Darley’s (2006) experiments, where people judge the moral responsibility of a person who killed another man by shooting him. When participants were told that the shooter performed the actions while under overwhelming duress (either they would be shot by a hijacker if they failed to comply or they were under a drug), they judged the man less responsible for the shooting. When participants were told that the shooter wanted to do it because the victim previously harmed him, the shooter was judged more responsible:
even when he performed that act in the presence of overwhelming external causation … that clearly made him do it! These findings suggest that people judge moral responsibility in a way that does not follow from a simple model of causal logic” (Wegner, 2008, p. 238).
It is not clear why Wegner says the external cause “clearly made [the shooter] do it.” It is clear to whom? Wegner’s description applies a normative ideal of an acceptable explanation where people should completely discount an internal cause when a compelling external cause precedes an event. Research shows that people may deviate from this ideal, but it does not follow that they are necessarily being illogical (McClure, 1998; Morris & Larrick, 1995). They may be applying a different causal logic to Kelley’s (1972) discounting principle. When people do not discount a cause in the presence of another cause, it is sometimes for good reason, not because they are illogical (McClure, 1998; Morris & Larrick, 1995). They may think that both causes affected the outcome.
A key issue for the exclusivity parameter is that both rare and common actions may have multiple causes (McClure, 1998). Furthermore, when judging explanations of intended actions, people prefer conjunctive explanations to single causes such as goals (Leddo, Abelson, & Gross, 1984; McClure, 1998; McClure, Lalljee, Jaspars, & Abelson, 1989). For example, people see a conjunction of the goal of wanting a meal and the precondition of having money as a better explanation of visiting the restaurant than either of the constituent causes. In addition, where two causes affecting an action are positively correlated, the presence of one cause renders the presence of the second cause more rather than less probable (McClure, 1998; Morris & Larrick, 1995). When a person decides to go on a trip, the presence of several preconditions for the action, such as an airplane and enough money being available, does not undermine people’s sense of will in undertaking the action. Nor does the presence of both internal and external causes, such as the desire to travel and a friend’s invitation to visit. This point applies to Woolfork et al.’s (2006) study described by Wegner, and applies when people have two motives to perform an action (McClure, 1998). If the two motives are positively correlated, the presence of one motive leads people to infer the presence of the other is likely, rather than discount it.
In evaluating whether people are discounting a cause, it is important to distinguish between the perceived causes of an action and the subset of causes given in explanations to other people. People’s communications often invoke a single cause to explain an action, but they nonetheless perceive multiple causes as necessary for the action to occur (McClure & Hilton, 1997, 1998). The cause in people’s communicated explanations is the most relevant or informative cause, given the causal background, rather than the sole perceived cause of the action. For common actions, such as buying a meal, people cite only the goal behind the action such as wanting food and omit causal preconditions such as money that they nonetheless see as necessary causes for the action. In contrast, with rare actions such as a poor person buying an expensive car, people’s explanation may cite the precondition that enables the unusual action and omit the relevant goal. Yet in both cases, they see goals and preconditions as necessary causes of the outcome (McClure & Hilton, 1997, 1998).
Given that actions typically have multiple causes and that this causal feature is recognized in social judgment, it follows that people may recognize intentionality or will even when other causes are present, rather than solely when other causes are absent, as proposed by Wegner. This is not to deny that intended actions may be attributed to other causes, particularly when external constraints on a person are strong or when people make excuses for undesirable actions. However, research fails to support the claim that people think that they intended an action only in the absence of other causes. The research on conjunctive explanations shows that people attribute intentions even when other causes affect an action.
Implications
This analysis points to a number of difficulties that arise in applying Michotte’s (1963) model of causal perception to explain will and intentions. Although the requirement that a cause must have temporal priority extends to intentions and will, intentional cognitions include premeditated intentions that occur well before the action, in contrast with Michotte’s studies on physical causation. A consistent relation between cognitions and actions is neither necessary nor sufficient for people to perceive their actions as intentional or willed. Many intentions are unfulfilled whereas other intentions are fulfilled by external events. There are also difficulties with the exclusivity parameter, as actions involve multiple causes and are perceived as such.
Michotte’s (1963) model applies to situations where a cause and apparent effect both occur prior to the perception. In contrast, with intentional actions, people experience a cognition before the action occurs and regardless of whether the action occurs. People also experience proprioceptive sensory feedback before and during an intended action, which differs from the observation of two events after an outcome (Blakemore, Frith, & Wolpert, 2001; Malle, 2006; Michotte, 1963; Nahmias, 2005). As Michotte (1963) noted, counter to Wegner’s extension of his theory of physical causation to will: “our voluntary activity is clearly different from similar activities of which we are observers, as happens in the case of the visual experiments” (p. 214) on physical causation. These observations do not contradict findings that actions are affected by unconscious factors. However, they do suggest that intentions and consciousness of will involve different processes from Michotte’s three parameters of causal perception (Smedslund, 2009).
It was noted earlier that some physical movements lead people to perceive intentional (social) causation (Heider & Simmel, 1944; Schlottmann et al., 2006). Could these movements explain people’s experience of will better than Michotte’s model? Is the issue that the perceptions underpinning physical and intentional causation are inherently different, or is it that Wegner overlooked the best set of physical movements to explain the experience of will? Research is needed to fully answer this question; however, the physical movements that produce perceptions of intentionality in physical events lack three features in people’s experience of their own intentions: intentions prior to actions; proprioceptive feedback during actions; and the capacity to initiate, modify, and cease actions.
Many of these features can be accommodated by the intervention model of causality (e.g., Gopnik & Schulz, 2007; Woodward, 2007). Whereas Michotte’s parameters parallel the sensory associations seen in classical conditioning, the intervention model extends operant learning, where organisms manipulate the environment. When a human sees the wind or a person shaking a branch and fruit drops, they don’t just salivate next time the wind blows; they also shake the branch (intervene) to get fruit (Woodward, 2007). Experiments show that people learn causal relations more rapidly and accurately when they can perform interventions and manipulate various candidate causes (Lagnado & Sloman, 2004). This enhanced learning occurs because interventions allow organisms to distinguish between competing sets of causal relations that are difficult to distinguish on the basis of correlational data such as Michotte’s parameters (Lagnado, Waldman, Hagmayer, & Sloman, 2007). People can also understand causal relations that exceed their own sensory experience, such as the causal mechanism that mediates between a branch moving and fruit falling. This also applies when causal mechanisms are invisible, as in concepts related to gravity; such relations pose a problem for the Michottian parameters that by definition apply only to visible variables (Woodward, 2007).
The intervention model also incorporates people’s capacity for counterfactual reasoning about different outcomes. This ability enables humans to plan and anticipate the consequences of different actions and means–ends options. People see their own actions as having the characteristics of interventions (Woodward, 2007). Thus this model can accommodate many features of intentional actions, such as distal intentions prior to actions, proprioceptive feedback during actions, and the capacity to modify actions. Woodward (2007) claimed that the intervention model also applies to causal reasoning about physical events.
Conceptual claims and assumptions
Conscious and unconscious causes
There are difficulties not only in Wegner’s research on conscious will, but also Wegner’s conceptual claims. Key issues emerge in the distinction between conscious and unconscious processes. In Wegner’s (2002) account, an unconscious event is an acceptable explanation for behavior, because the “real” causes of actions are unconscious. In contrast, if an event is conscious, this is described a priori as an unacceptable or unscientific explanation. Wegner (2008) writes that that “our behavior is caused by mechanisms of mind and not by our free willings” (p. 236).
One issue with this argument is that events that scientists currently classify as conscious and unconscious are aspects of the same information-processing system (cognition) that is performed by the same mechanism (the brain). If neuroscience could give people access via an electrode to unconscious events “causing” their actions, such as relevant cells firing, would the claim hold that the person is not aware of those causes of their behavior? In principle if not yet in practice, neuroscience could make these connections. It is possible that the brain includes a natural link analogous to this constructed link between unconscious and conscious events.
Nahmias (2005) observed that Wegner fails to show that conscious intentions have no input into the system that produces actions. Malle (2006) added to this point in noting that it is illogical for researchers to allow that conscious cognitions may be effects in causal chains while denying that these cognitions could function as causes. Either conscious cognitions are not events at all, or as events with causes they have the potential to causally influence other events (see also Bayne, 2006; Holton, 2004). In response to Wegner’s claim that people lack access to the “real” unconscious causes of their behavior, Nahmias (2005) noted that:
a lack of direct introspection of what happens below the surface does not entail that our experiences are illusory any more than our lack of direct perception of the molecular activity that happens below the surface when one billiard ball strikes another entails that our experience of causation in that case is illusory. (p. 777)
Wegner’s account, like that of some other researchers, shows reluctance to include conscious cognitions in an account of behavior (Kihlstrom, 2008). Yet conscious cognitions may have evolved because they represented an adaptive link to cognitive causes of actions, albeit an imperfect route. A theory of consciousness needs to explain why related mechanisms such as decision-making and premeditation evolved and may have adaptive functions. It is implausible that these mechanisms evolved primarily as illusion-producing mechanisms to mislead people about reality.
Wegner (2008) proposed that the illusion of will had evolutionary value in signaling our ancestors’ intentions to others so that the intentions could be countered. The consciousness of will may equally have served to conceal people’s intentions from others, so that the intention would be realized rather than thwarted. For example, human ancestors may have concealed their intentions from others when they hid food (Byrne & Whiten, 1988). In either case, it is unlikely that consciousness of will would be adaptive if it was illusory. Prior intentions contribute to successful planning and enable people to anticipate and respond to distal events.
There is no question that unconscious factors affect judgments, preferences, and behavior. But it is arbitrary to claim that these are the only cause or the sufficient cause of behavior. Unconscious processes are part of a causal network of events that includes environmental events such as priming, unconscious processes, and conscious cognitions. Environmental events may in turn be artifacts of conscious intentions, as in an experimental design for which the designers may rightly claim authorship. Kihlstrom (2008) claimed that research shows that controlled processes exercise a significant influence on judgments and actions that differs from the automatic processes that many social psychologists emphasize. Suhler and Churchland (2009) note that evidence that people are influenced by unconscious factors is counterbalanced by evidence that people exercise control over their goals and resist distraction from environmental events. This action control operates at conscious and unconscious levels (Kandel, 2006).
Some recent research on intentions differs from classic formulations of will in allowing that intentions and goal pursuit are not always conscious—people may have intentions that they are unaware of. Nahmias (2005) proposed that intentions extend to automatic behaviors, and that people often feel the authors of well-rehearsed automatic behaviors, such as sporting or musical activities, in the absence of conscious intentions to perform the actions. The state of flow occurs when people exercise the skills to perform challenging activities without awareness (Csikszentmihalyi & LeFevre, 1989).
The phenomenal will and the empirical will
Questions also apply to Wegner’s (2002, 2004) distinction between the phenomenal will, which is described as illusory, and the empirical will, which is seen as a valid explanation (Bayne, 2006). The phenomenal will is the person’s experience of will, whereas the empirical will is defined as “the causality of the person’s conscious thoughts as established by a scientific analysis of their covariation with a person’s behavior” (Wegner, 2002, p. 14). Wegner (2002, 2004) mixes this specific definition of the empirical will as a conscious process with broader definitions, defining the empirical will as the relationship between mind and action (2002, p. 15) or between thought and action (2004, pp. 681, 683). Wegner (2004) also claimed: “At its most general, willful behavior can be said to occur whenever there is evidence that information input to a system caused a change in the subsequent behavior of the system” (p. 681). It is not clear how this broad definition distinguishes will from the response of any organism to stimuli or feedback, so it is unclear why Wegner retains the concept of empirical will.
Wegner (2004) chided his critics for failing to recognize his distinction between the empirical and phenomenal will and claimed to be “surprised that … commentators would mistake my meaning” (p. 683). However, Wegner’s model does not clarify the relationship between the two concepts. The differing definitions of the empirical will do not clarify what exactly the empirical will is and how it differs from the phenomenal will, or why evolution should have created two will mechanisms with no connection. The empirical will does not appear in Wegner’s diagram of his model (Wegner, 2002, p. 68; Wegner & Wheatley, 1999, p. 483). This key aspect of the theory requires clarification.
Volitional concepts: Superstition or science?
Wegner (2008) argues that teleological concepts such as will and intention involve magical or supernatural thinking, and that the concept of will is like a magic spell. This image invokes Dennett’s (2006) description of beliefs in supernatural agents such as gods in the language of a spell. In extending this “spell” language to human volition, Wegner implies that he has cleared the universe not only of agentic gods but also of agentic humans. Mele (2008) objected that there is no need to motivate a skeptical analysis of intentionality “by importing outlandish ideas to debunk: for example, the idea that supernatural, magical selves cause intentional actions” (p. 343). Malle (2006, p. 208) described the skepticism about intentionality as a battle against windmills, where scholars fight perceived enemies of the scientific world-view that in reality are not enemies. Dennett (2003) accepted Wegner’s evidence that people make errors in their understanding of their own willed actions, but in contrast with Wegner, he argues in evolutionary terms that language and conscious processes enable a higher level of voluntary action that other species lack, such as being able to ask others about their intentions. He claims: “It is this kind of asking, which we can also direct to ourselves, that creates the special category of voluntary actions that sets us apart” (p. 251; see also Dennett, 2008).
Wegner (2002) claimed that the concept of will implies a homunculus or explanatory entity that cannot itself be explained. However, as he acknowledges elsewhere (Wegner, 2005), this is not necessarily the case, as volitional concepts can be given operational definitions. Wegner (e.g., 2002) uses the terms “conscious will” and “intention” interchangeably, and intentions and other volitional concepts have been examined scientifically in several theories, including action control theory, decision-making, scripts, and self-regulation (e.g., Baumeister, Hetherington, & Tice, 2004; Gollwitzer, 1999; Haggard & Clark, 2003; Malle, 2006; Schank & Abelson, 1977; Vallacher & Wegner, 1987). This research refutes the claim that teleological concepts are incompatible with scientific accounts of behavior (Baumeister, 2008).
These theories do not portray intentions or will as mystical, but show how volitional cognitions relate to actions and how self-regulation is either strengthened or depleted (e.g., Baumeister et al., 2004). They also clarify how people generate conscious intentions and what factors enhance or impede the realization of those intentions. For example, Gollwitzer showed that when people consciously form implementation intentions, they accomplish more of their goals. Research on self-initiated action is clarifying the relation between conscious cognitions and actions (e.g., Haggard & Clark, 2003; Moore et al., 2009), and the interplay of controlled and automatic processes (Kihlstrom, 2008; Suhler & Churchland, 2009). These findings challenge the view that conscious intentions lack efficacy and are not a part of the causal chain leading to actions (Mele, 2009).
Research in social cognitive neuroscience has shown that people’s inferences about intentional actions involve different brain substrates from inferences about physical events (Decety & Grezes, 1999; van Overwalle, 2009). Intentional actions trigger automatic intention inferences; when humans (and monkeys) register a human movement, they automatically infer a corresponding goal (Allison, Puce, & McCarthy, 2000; Keysers & Gazzola, 2007; Wood & Hauser, 2008). This process has adaptive value in enabling rapid responses to other animals’ actions. This research relates to people’s sense of their own movements, in that mirror neurons activate with both the perception and the performance of the same actions (Fogassi et al., 2005; Rizzolatti & Craighero, 2004). However, non-human primates can infer the intention of a human action such as throwing a rock even though they are unable to perform the action (Wood & Hauser, 2008). Beauregard (2007) argues that neuroimaging studies on self-regulation and therapy show that voluntary actions and cognitions affect brain activity. This neuroscience research requires recognition of people’s consciousness of intentions as cause as well as effect.
These findings suggest that cognitions involving our own intentions and actions have several features. First, they may involve different brain processes than perceptions of physical causation. Second, they give equal weight to cognitions that precede actions and retrospective observations of actions. Third, they include the capacity to intervene in causal relations and respond to proprioceptive feedback during movements. Fourth, they embed intentions in models of multiple causation that recognize that each action and event results from a range of causes in causal chains. Fifth, they acknowledge causal processes at different levels of analysis; they reject a selective reductionism proposing that unconscious events are the “real” cause of actions and, on the one hand, do not further reduce these causes to chemico-physical events and, on the other hand, preclude higher levels of causation. Sixth, they propose that consciousness may have evolved due to its adaptive functions rather than being a misleading or epiphenomenal device. Seventh, they recognize a continuum of efficacy, from an inability to move one’s finger to grand actions such as circumnavigating the planet or writing a book at will.
These observations underscore two points. First, these theories provide coherent, well-researched alternatives to Wegner’s account of conscious will, so there is no need to accept Wegner’s Michottian model, which has significant weaknesses, because nothing else is on offer. Moreover, by allowing that people’s actions may be affected by conscious intention (“will”), these alternative accounts allow for a continuum of efficacy where, at one extreme, people generate and accomplish intentions and, at the other, people cannot form simple intentions or accomplish goals. It may be useful to integrate elements of Wegner’s account with these alternative accounts, and recent research on intention and action may assist this linkage (Haggard & Clark, 2003; Moore et al., 2009; Suhler & Churchland, 2009). This is important, because no other theory has systematically connected will with theories of self-attribution.
Sometimes Wegner (2002) distinguishes his theory of conscious will from free will (e.g., pp. 318–22), whereas elsewhere he treats them as synonomous (e.g., p. 26). It is debatable whether the concept of “free will,” which is a lay and philosophical concept, is useful in scientific research on intentionality and volition (Malle, 2006). Many models of action avoid the term “free will” (e.g., Haggard & Clark, 2003), as do lay theories (e.g., Malle, 2004, 2006; McClure & Hilton, 1998). Research framed in terms of intentions and self-regulation makes similar assumptions about human action to those implicit in the notion of free will, while offering more precise concepts and definitions. So the retention of the term “free will” may not be a central issue.
Conclusion
This article is not claiming that psychological science is unable to give a causal account of will or intentionality. Nor is it arguing that people are accurate in their perception of the causes of their behavior. People are unaware of many unconscious factors affecting their behavior and, although people may be conscious of their intentions, their reported motives for actions are often inaccurate (McClure, 1983, 1991). What this article is arguing is that Wegner’s use of Michotte’s (1963) model of causal perception to explain conscious will is not a convincing account. The data do not justify the conclusions, and several concepts underpinning the research are problematic. Other theories in psychology and neuroscience provide an alternative framework that recognizes the boundaries of self-efficacy but allow that conscious cognitions may be causes as well as effects. Theories that link the consciousness of intention to attribution processes need to recognize that each action and event reflects a network of causes rather than one cause.
Footnotes
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
