The Where and When of Intention

See allHide authors and affiliations

Science  20 Feb 2004:
Vol. 303, Issue 5661, pp. 1144-1146
DOI: 10.1126/science.1095331

Intention is judge of our actions.

-Michel de Montaigne

At a moment of your own choosing, snap your fingers. Now ask yourself: “When did I first feel the urge—or intention—to make that snap? Was it a full second before my fingers moved?” Although that duration might seem counterintuitive, human brain studies using electroencephalography (EEG) have long suggested that some part of your brain was already moving toward that decision well before you were aware of it. Spontaneous, voluntary movements are preceded by a progressive rise in motor area activity known as the readiness potential (RP) (14) more than a second before you make your move (see the figure). Although we are subjectively unaware of this buildup of activity, does this mean that we are not aware of anything before our fingers suddenly jerk into motion? Or do we have some sense that we are about to act, some notion of intention just before our bodies begin to move?

Time course of a voluntary act.

Subjects asked to report when they first felt the “urge” or intention to move typically report the time as ∼200 ms before the actual movement. (Top) EEG studies have shown that a progressive rise in neural activity (the readiness potential) precedes the subjective intention by more than a second, and is larger when subjects judge the time of their urge to move (red trace), rather than the movement itself (black trace) (16). (Bottom) Bringing this tradition to neuroimaging, Lau et al. (6) found increased blood flow in the dorsal prefrontal cortex, the intraparietal sulcus, and the pre-supplementary motor area when subjects attended to the time of their urge to move, rather than the movement itself. [Graph adapted from (16)]


To explore this issue, one set of early experiments asked participants to make a spontaneous finger movement—at a time of their choice—while watching a spot moving around a clock face. Subjects were asked to report the time at which they first felt the urge to move. Their typical answer: ∼200 to 250 ms before the time of their actual movement (5). This experimental design has had a long and often controversial history—after all, how do we know subjects aren't simply attending to the beginning and end of the same movement, or deciding that the time of their intention logically must precede the time of their action? Given these uncertainties, it has remained unclear whether the urge to act, and the action itself, represent actual differences in brain states. Onto this stage enter Lau et al. (6), on page 1208 of this issue, with a functional magnetic resonance imaging (fMRI) experiment that directly addresses this question.

In Lau et al.'s study, participants made a spontaneous finger movement and reported the time at which they first felt aware of the intention to move (I-condition) or they actually moved (M-condition). In line with previous findings, subjects reported the urge to move an average of ∼200 ms before the time of the movement. In this study, however, the goal was to investigate where, not simply when, brain activity occurred during the I- and M-conditions. The authors used neuroimaging to achieve this goal, assuming that the two conditions represented attending to either the intention or the movement.

Modulating neural activity through attention is a powerful tool for neuroscientists. Many studies have shown that paying attention to a sensory stimulus increases the BOLD (blood oxygenation level-dependent) signal in the corresponding sensory part of the brain (7). For example, even if a retinal stimulus remains the same, attending to it will increase blood flow in the visual cortex. The attentional spotlight is a valuable tool not only for studying representations of external stimuli, but also for looking inward—for example, by attending to different internal representations in working memory (8). To that end, recent fMRI studies on movement have compared brain states during differing degrees of attention to a self-made motor act (9, 10) or during spontaneous versus cued action (11, 12). Taking the next step in this tradition, Lau et al. bypassed several potentially confounding factors by using a single action in both conditions: a single, voluntary lift of the finger at a time of the subject's choosing. Subjects attended to different aspects of the same act—the urge to begin, or the action itself.

The blood flow data showed greater activity in three brain areas during the I-condition. One of these, the pre-supplementary motor area (pre-SMA), is known to become active when subjects voluntarily generate movement (13), even during simulation of movement without actual execution (14). Furthermore, stimulation of the neighboring SMA in humans reportedly generates an “urge” or “anticipation” of movement (15). Lau et al. report that the pre-SMA BOLD signal peaks at ∼3 s after the keypress, leading them to argue that neural firing preceded the motor act by some 2 to 3 s (because the hemodynamic response takes 5 to 6 s to peak). Does the pre-SMA activity reflect the readiness potential? This seems generally consistent with EEG findings suggesting that attending to intention increases the readiness potential (16). However, note that Lau et al.'s pre-SMA activity is midline, whereas the readiness potential in (16) is lateralized.

Next, Lau et al. found increased activity in the dorsal prefrontal cortex (DPFC), a key structure implicated in generating and developing plans for voluntary action (17, 18). Finally, they observed that the I-condition engendered higher activation in the intraparietal sulcus (IPS), which is among the areas most consistently activated by movement preparation (19) as well as attention to stimulus attributes (7). Because patients with parietal lesions have no deficit in making a willed action (16, 20), this area may be involved in self-monitoring actions rather than forming them (21).

When viewed in combination with other studies on attention to action, one hypothesis could be that the DPFC is involved in generating the intention to move, whereas areas in the parietal lobe, and perhaps also the pre-SMA, begin to simulate or anticipate future movement. Although Lau and co-workers suggest that the pre-SMA activity “reflects the representation of intention,” the full story should include at least the parietal lobe as well. Recently, Sirigu et al. found that patients with parietal lesions showed no distinction between their time estimates in the I- and M-conditions, whereas normal subjects (and cerebellum-lesioned controls) consistently answered that their intention preceded their action by ∼250 ms (16). This suggests that parietal patients have an undamaged ability to time their movement, but damaged access to an internal model that simulates future activity. One must draw parallels between studies cautiously, however, as the parietal activation reported by Lau et al. was anatomically much more dorsal than the lesions in the Sirigu et al. study. Because pre-SMA and parietal areas both seem to be important for representing intention, future research will need to clarify the relationship (causal? parallel?) between them. One way forward would be to revisit the I- and M-conditions in patients with pre-SMA lesions. Future investigations might go further by combining into one study voluntary and unpredictably forced actions, or by imaging patients with schizophrenia, which is characterized by problems with overattribution (that is, patients assume intention for actions that are not their own) (22).

The new study suggests a deeper question: to what, exactly, are the subjects attending in the I-condition? As mentioned, one theory postulates that the I-condition forces subjects to access an internal model of the desired movement (16). The idea is that during a self-generated action, copies of the commands sent to the muscles are fed into a predictive (or forward) internal model (23), whose job it is to simulate what is expected next. But some questions arise. First, if the readiness potential represents an internal model that begins “revving up” at least a second before the motor action, why is this model only accessible to awareness ∼200 ms before the action? Does the model need to reach a certain threshold of activity to be accessible, and, if so, what is special about this threshold, neurally speaking? Second, do we consciously experience the urge to move if we are not asked the question? That is, does attending to the subjective intention generate it? Third, in the Sirigu et al. study, patients with lesions of the cerebellum (thought to function as an internal model) (24) performed like normal subjects under both the I- and M-conditions (16), suggesting that certain internal models may not be accessible to awareness. If there exist several internal models in the brain, what is the neural difference that makes some accessible to awareness and others not?

The finding that there are distinguishable brain states between the I- and M-conditions reinvigorates discussion about intentionality, but the final interpretation of these results lies in a thicket of further questions. Most broadly, it remains to be understood how the neural events are related to the phenomenal experience that “I” was the author of an action. The internal model hypothesis suggests this relationship may be due to matching the consequences of a movement against its internally predicted effects. But predictability cannot be the complete story, because people judge the time of their own actions and the actions of others equally well—but strangely, the “actions” of a nonbiological machine are judged quite differently, even when they are visually identical and equally predictable (25). This suggests that intentionality might even be judged retrospectively, an illusion arising from watching yourself (or another agent) make actions (26). This is consistent with the idea that you represent the actions of others by analogy with your own, inferring their intentions by watching their actions. Thus, in contrast to Montaigne's belief, it may be that action is the judge of intention.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
View Abstract

Stay Connected to Science

Navigate This Article