The Illusion of Conscious Will (35 page)

This was exactly what Wegner and Fuller (2000) found in the muscle reading investigation. Participants in an experiment who were instructed to apply upward pressure for the entire muscle-reading session— presumably pushing up against the fingers of the confederate that were resting atop their own—went beyond the usual 87 percent level of accuracy to achieve 94 percent correct, a significantly higher rate. It may be, then, that part of the process whereby correct answers are provided on behalf of the communicator involves the production of an ironic automatic behavior. The facilitator’s attempt
not
to put words into the communicator’s mouth ends up doing exactly that.
⁴

This experiment also tested how participants reacted when they were told to go ahead and answer the questions for the communicator. A blatant instruction to “do it for them” led to an even higher level of correct answering on the easy questions—an increase to 96 percent correct. And, indeed, when these participants were asked whether they followed the instructions to make the responses for the communicator, they were more inclined than participants given the usual muscle-reading instructions to agree that they had taken over and made the responses. Yet their level of attribution of the responses to the communicator persisted essentially unchanged (39 percent). Through the rigmarole of touching the other person’s fingers and waiting for their possible (but imaginary) muscle movements, participants became convinced that the communicator must have had some effect on the answer even when they themselves were admonished to provide it.

4.
This is not the whole story on how this automatism is caused, of course, and more research is needed to uncover the process precisely. It makes sense, for example, to think that in the case of normal college students who are “facilitating” for other normal college students, there is a strong suspicion by the facilitator that the communicator does indeed know the answers to the easy questions. In fact, this suspicion could make it somewhat strange and embarrassing not to answer correctly on behalf of the communicator, especially when one’s answers are known to the communicator. This is not the entire solution to the problem of how the correct answers are given either, however. Wegner and Fuller (2000) conducted a series of tests that appear to undermine the likelihood of this interpretation. In their studies, people who were asked to read muscle movements of someone who was ostensibly given questions
subliminally,
and who thus could not know whether the facilitator was answering correctly on their behalf, nonetheless gave a high proportion of correct answers to easy questions.

How is it that participants come to attribute a substantial proportion of their correct answers to the communicator? All that seems to be necessary for such attribution is the belief that it is possible. In our experiments, we regularly measured belief by asking people to indicate whether they believed that they could discern another person’s preferences by reading the person’s muscle movements. Such belief was very strongly predictive of attribution of the answers to the communicator in all the studies. Of course, it is unclear when there is a correlation of this kind what is causing what. Although it might be that belief in muscle reading causes the attribution of action to the communicator, it could also be that the attribution of action to the communicator causes belief in muscle reading.

To ascertain whether belief was indeed causal, in one study we directly manipulated belief. For this experiment, the parallel between facilitated communication and our muscle-reading task was emphasized to participants. Then participants watched a video of a segment of a PBS
Frontline
program entitled “Prisoners of Silence” (Palfreman 1993). This program began with the history and apparent promise of FC and concluded with a review of the more recent findings indicating that the technique is invalid. For our study, some participants saw the breathless buildup of FC only, whereas others saw the entire episode, complete with debunking. We found that these groups then differed in their impressions of what had happened in their own muscle-reading sessions. Both groups showed similar levels of correct answers to the easy questions, but the pro-FC video group attributed their performance more strongly to the communicator (38 percent influence) than did the anti-FC video group (24 percent influence). It is interesting that even this latter group did not give up on the idea of muscle reading entirely, as they maintained a tendency toward action projection even after having seen a strong argument against it. However, it is clear that undermining belief in the possibility of outside agency indeed hampered action projection in this situation.

These findings suggest that people might conceivably project actions to others whenever they believe that others are potential sources of those actions. When else does this happen? Consider any instance of helping. A parent helps a child get dressed and may easily come away thinking the child put on the mittens successfully and will be able to do it again. Next time, though, the parent is struck with the child’s apparent lapse in memory. It seemed as though the child was perfectly capable of doing this independently but now can’t do it. Perhaps the parent was projecting the action in the initial session and ended up attributing to the child what the parent had actually accomplished.

And what about a teacher helping a student with fractions? The student is trying to figure out what 1/3 of 5/12 might be. The teacher produces an elegant explanation of the problem, the student correctly answers all the subquestions the teacher volunteers during the solution (“And how much is 3 times 12?”), and the final answer pops up just as it should. The teacher may well feel that the student has done the problem and should now be able to do others. Time to drift off to the teachers’ lounge? The real test comes, though, if the teacher remembers to say “now you try one.” All too often the student turns directly into stone or some other dense material. The action of solving the problem was projected by the teacher onto the student in the first place, and the assumption that the student had learned turned out to be too optimistic.

Yet other examples of action projection arise not with children or students but in the pursuit of sexual partners. One person romancing another, for example, may become very ready to presume that the interaction is proceeding as hoped. The person who is the target of these advances may be perceived to have reciprocated, or even to have originated them, largely through a process of action projection. The finding that men tend to read sexual meanings into women’s behavior (Abbey and Melby 1986) suggests that action projection could occur in such situations, with the result that men might come away from romantic interactions believing that women had acted far more provocatively and sexually than was actually the case. This process could lead to a striking conflict of perceptions, in which the male self-righteously claims how much the female “wanted it,” while the female validly reports that the male overstepped her clearly indicated bounds.

Action projection is a process in which one person operates another almost like a puppet and then attributes life to the puppet. Each of us is potentially a Geppetto with our own band of Pinocchios. The process of action projection is exceedingly difficult to notice, however, so the act of puppetry will not easily be detected. People who make horses do math and who procure messages from victims of autism when those victims can’t even talk seemingly do all this without much clue that their own actions are creating the exhibition. Action projection may well operate pervasively in many situations, but it is unlikely when we are watching this show that we will concurrently be keen enough self-observers to realize that we are the ones pulling the strings.

The Whodunit Problem

It appears that people can get pretty bollixed up in their understanding of who did what in a social interaction. In a way, this is inevitable. Even with the computational tools of the average rocket scientist, for instance, it could be a sizable task to figure out who did what in just half an hour of a typical facilitated communication interaction. There are all kinds of behaviors going on, as well as hints about who was thinking about what, not to mention information about the priority, consistency, and exclusivity of every potential pairwise thought-action combination in both partners. Imagine what this would look like if we were to try figuring out who willed what actions by whom in the course of a few hours of court proceedings or in the snappy repartee of a good romantic comedy. Whew.

You or Me?

People must necessarily use a shorthand approach, a way of estimating who is willing what in the heat of social interactions. One such approach is simply to assume that the predominant willing is coming from one of the people and then to sort things out from there. Which person gets picked for this role could be largely a matter of basic principles of attention (Taylor and Fiske 1978). The person who is louder or taller or moving faster or wearing the brighter clothing could draw the viewer’s attention and so could become the default source of the perceived will in the interaction. In fact, there is evidence that such standout people are often perceived to be running the show, even if they are not actually in charge (McArthur and Post 1977). If this salient person is oneself, for some reason (if one is particularly self-conscious, for example), then the focus on oneself will promote a tendency to default to self when questions of who did what arise (Duval and Wicklund 1972; Storms 1973).

The attribution of action to oneself can be enhanced unconsciously by the presence of subliminal stimuli that direct attention to self. Participants in experiments by Dijksterhuis, Wegner, and Aarts (2001) were asked to react to letter strings on a computer screen by judging them to be words or not and to do this as quickly as possible in a race with the computer. On each trial, the screen showing the letters went blank either when the person pressed the response button, or automatically at a short interval after the presentation (about 400-650 milliseconds). This made it unclear whether the person had answered correctly and turned off the display or whether the computer did it, and on each trial the person was asked to guess who did it. In addition, however, and without participants’ prior knowledge, the word
I
or
me
or some other word was very briefly presented on each trial. This presentation lasted only 17 milliseconds and was both preceded and followed by random letter masks such that participants reported no awareness of these presentations.

The subliminal presentations influenced judgments of authorship. On trials with the subliminal priming of a self-relevant pronoun, participants more often judged that they had beaten the computer. They were influenced by the unconscious priming of self to attribute an ambiguous action to their own will. In a related study, participants were subliminally primed on some trials with the thought of an agent that was not the self— God. Among those participants who professed a personal belief in God, this prime reduced the attribution of the action to self. Apparently, the decision of whether self is the cause of an action is heavily influenced by the unconscious accessibility of self versus nonself agents. This suggests that even when we are not thinking consciously about ourselves or others, we may be influenced by circumstances we can’t discern to judge the pre-dominant authorship in one direction or the other.

One such nonobvious influence on the computation of self’s versus other’s will is the psychological
punctuation
of the interaction. We all remember examples from grammar class in which punctuation can change or even reverse meaning (compare “John said Mary is oversexed” with “John, said Mary, is oversexed”). A typical social interaction involves a series of actions by two or more persons that are interleaved over time and that therefore may present different meanings depending on a sort of punctuation as well (Whorf 1956). With repunctuation of the interaction, for example, a bright young lab rat might reflect, “I have got my experimenter trained. Each time I press the lever he gives me food” (Bateson and Jackson 1964, 273). More generally, any interaction in which persons A and B alternate their behaviors in sequence might be punctuated as “A causes B’s act” or “B causes A’s act,” and it may be difficult or impossible to tell from the sequence itself which punctuation is more apt.

A classic example of this phenomenon is the viciously circular marital problem recounted by Watzlawick, Beavin, and Jackson (1967). A couple is engaged in a constant relationship struggle in which he passively withdraws while she nags and criticizes. In explaining the situation, the husband is likely to describe his withdrawal as a
response
to her nagging, whereas the wife is inclined to describe her nagging as a
response
to his withdrawal. Their fights thus consist of a monotonous exchange of “I withdraw because you nag” and “I nag because you withdraw.” In fact, both things are happening, but it is difficult for either partner to recognize the whole sequence or its repetitive self-sustaining quality. Instead, each believes that his or her own behavior is quite involuntary, occurring largely because of the other’s prior behavior.

Such gridlocked interactions have been explored in laboratory experiments by Swann, Pelham, and Roberts (1987). For these studies, college students were paired for interactions in which they played roles of world leaders during the Cold War. Each “leader” was given a series of statements about nuclear weaponry from which to choose, and the two exchanged statements six times over a phone hot line. The statements varied from warlike (“We will not tolerate any threats to our national security”) to conciliatory (“Our goal is to establish a climate of mutual trust”). Some of the participants were instructed to adopt an “offensive” set for the exchange in that they were to convince their partners that they were very powerful leaders who would initiate a first strike with nuclear weapons if conditions made it necessary. Other participants were given a “defensive” set and asked to decide whether they would be willing to initiate a nuclear first-strike against their partners based on the statements their partners made during the interaction.