Elsevier

NeuroImage

Volume 202, 15 November 2019, 116140
NeuroImage

Latent awareness: Early conscious access to motor preparation processes is linked to the readiness potential

https://doi.org/10.1016/j.neuroimage.2019.116140Get rights and content

Highlights

  • An RP-like signal is directly related to awareness of intention in voluntary action.

  • People can control behaviour based on awareness of intention to act.

  • Motor preparation processes start before increases in visual attention.

Abstract

An experience of intention to move accompanies execution of some voluntary actions. The Readiness Potential (RP) is an increasing negativity over motor brain areas prior to voluntary movement. Classical studies suggested that the RP starts before intention is consciously accessed as measured by offline recall-based reports, yet the interpretation of the RP and its temporal relation to awareness of intention remain controversial. We designed a task in which self-paced actions could be interrupted at random times by a visual cue that probed online awareness of intention. Participants were instructed to respond by pressing a key if they felt they were actively preparing a self-paced movement at the time of the cue (awareness report), but to ignore the cue otherwise. We show that an RP-like activity was more strongly present before the cue for probes eliciting awareness reports than otherwise. We further show that recall-based reports of the time of conscious intention are linked to visual attention processes, whereas online reports elicited by a probe are not. Our results suggest that awareness of intention is accessible at relatively early stages of motor preparation and that the RP is specifically associated with this conscious experience.

Introduction

Voluntary actions are often defined as those actions that are not triggered by an external stimulus, but are rather initiated endogenously (Haggard, 2008; Passingham et al., 2010). Such ‘self-paced’ actions have been consistently shown to be preceded by the Bereitschaftspotential (BP, “readiness potential”, RP), a slowly increasing negativity over the motor cortex. The RP was first described by Kornhuber and Deecke, (1965), and was later famously claimed to precede the conscious experience of intention in voluntary action (Libet et al., 1983). Libet asked participants to note the time on a clock when they first became aware of their intention (“urge”) to make a movement. This experience occurred on average 206 ms prior to action, while the RP itself began much earlier, often over 1 s before movement (Libet et al., 1983). This finding implied a very limited role for conscious control in voluntary action. However, this conclusion remains controversial. Both the method used to estimate the time of awareness of intention and the assumptions made about the RP itself have been challenged (for a review, see Guggisberg and Mottaz, 2013).

In the original Libet paradigm the estimation of the conscious experience of intention is only reported ‘offline’. Participants noted the time at which they became aware of their intention to move, but reported this only after they actually executed the action. This method 1) allows for the possibility of postdictive reconstruction – i.e. people may not have any genuine experience of intention before the movement, but might simply infer that they must have intended to act because they did act (Wegner, 2002), and 2) assumes that people can only consciously access their preparation for the impending movement once this has reached a specific, fixed threshold.

However, intention to move may develop progressively, rather than appearing suddenly (Matsuhashi and Hallett, 2008). In the unusual setting of the Libet experiment, the action is not motivated by any reason or constraint, nor does it bear any consequences in the external world beyond its own execution. Thus, it may not make sense to spontaneously access motor preparation information at early stages. However, in real-life scenarios where information about motor preparation is crucial for guiding behaviour, a stronger experience of preparation at earlier stages may be highly functional.

Here, we have investigated conscious intention in situations where ongoing motor preparation is relevant for task performance. In a previous study (Matsuhashi and Hallett, 2008), participants were instructed to perform self-paced movements. A computer occasionally played a sound, whose timing was unpredictable. If at the moment of the sound participants felt they were already preparing to move, they were told to inhibit this movement and wait for at least 5 s before moving again. That is, their decision to inhibit or not depended on their ‘online’ awareness of ongoing motor preparation. By studying the distribution of keypresses with respect to tone onsets the authors estimated the time of awareness as being 1.42 s before action, which was notably earlier than Libet’s 0.2 s. These results are consistent with a low threshold of motor preparation for latent awareness of intention, and imply that this ‘latent’ awareness of intention may be part of flexible action control.

In line with this study, we hypothesised that people do have some conscious access to their motor preparation processes before action. Further, we hypothesize that although they may normally experience intention to move only at late stages (when some ‘spontaneous threshold’ is exceeded), the intention is potentially accessible at earlier stages (‘latent threshold’) and can be used as a basis for action decisions. Our participants performed self-paced key presses while viewing a letter stream (cf. Soon et al., 2008). An ‘interrupting’ cue was inserted into the letter stream at random. Participants were instructed to respond to the interrupting cue if they felt they had already begun preparation of their next movement, and to ignore it if they were not preparing to move (see methods below). Thus, the cue would sometimes intercept voluntary motor preparation before an action was executed, and sometimes not. This online probing method allowed us to 1) interrupt motor preparation at various stages, 2) make real-time awareness of intention relevant for task performance and 3) eliminate the possibility for reconstruction by turning the action itself into a report.

Thus, the participants’ experience of their own motor preparation determined how they should respond to the cue. If our hypothesis is correct and the RP is somehow related to participants’ conscious awareness of motor intention, we should observe differences in brain activity prior to the interrupting stimulus as a function of how participants responded to that stimulus. Instead, if participants have no real-time experience of their motor intentions at all (as strongly postdictive theories suggest), their responses to the cue could only be random guesses. No systematic differences would then be expected in neural activity prior to the interrupting cue between cues where participants report an intention to move, and cues where they do not.

While Matsuhashi and Hallett (2008) recorded EEG during their experiment, there was no overt behavioural event allowing them to distinguish whether people did or did not have a conscious intention to move at the moment of any given tone onset. Their participants might have waited for >5s before the next movement because they actively inhibited a movement in response to the beep or simply because they happened to not feel any urge to move during the 5 s period. Thus, they were not able to identify the neural correlates of conscious intention. By using a “contingent go” instead of a “contingent veto” instruction, our study included a behavioural marker that could classify each probe event as either coinciding or not coinciding with a conscious intention.

Crucially, the sampling strategy of our paradigm differs from most RP studies. Classical RP studies lock EEG data to voluntary movements. They reveal neural activity prior to action, but ignore neural activity at other times. This biased sampling means that the RP pattern might not be specifically related to voluntary action preparation, nor to awareness of intention (Mele, 2011). In particular, RP-like fluctuations might also occur in the absence of actions and conscious intentions, but those RPs would be invisible to the action-locked methods used in classical RP paradigms. Recent computational models showed that simply averaging an ongoing stochastic signal time-locked to a threshold crossing event can reproduce the form of the RP (Schurger et al., 2012; see also Murakami et al., 2014). Since the underlying fluctuations in such models are continuous, RP-like forms should also occur even in the absence of action.

The current study investigates the specificity of the RP using a method that avoids this particular bias. We interrupted ongoing EEG with a random cue. If RP-like signals simply occur as part of ongoing stochastic fluctuations and are unrelated to awareness, they should be equally visible whether the cue interrupts an intention to act and when it does not. On the other hand, if the RP is specifically related to the participants’ experience of intention, we should see some RP-like signal prior to interrupting cues where participants reported a feeling of intention, but not otherwise.

A final concern about Libet-type studies of intention relates to perceptual attention and the time of awareness. Libet’s participants reported the position of a rotating clock hand after every action. Several authors have speculated that the visual processing of the clock might not be independent of either the actions made or of the experiences reported (see ‘Open Peer Commentary’ and ‘Author’s response’ for discussion in Libet, 1985). To the best of our knowledge, however, the three-way relation between visual attention, motor preparation and conscious intention has not been directly investigated.

In our study, participants viewed a letter stream at constant frequency (similar to Soon et al., 2008), rather than the more familiar rotating clock (Libet et al., 1983). After executing an action, participants were sometimes asked to report the letter that was on the screen when they first felt an intention to move, a recall-based mental chronometry method analogous to Libet’s one. The kind of periodic stimuli we used elicits steady-state visual evoked potentials (SSVEP) at the stimulation frequency and its harmonics. SSVEP amplitude reflects the variations in visual pathway processing gain with visual attention. For example, switches of attention between two stimuli of different frequencies can be tracked by analysing fluctuations of SSVEP amplitude at the corresponding frequencies (Müller et al., 1998). Canonical correlation analysis (CCA) between EEG and a target frequency can be used to study the allocation of visual attention (Lin et al., 2006). In typical BCI applications, multiple stimuli flicker at different frequencies. The frequency that has the highest CCA score is interpreted as selected in attention (e.g. Chen et al., 2015). Our study used a single stimulation frequency, corresponding to the letter stream presentation, and we used CCA analysis to study how attention to the letter stream was modulated over time. In particular, we used canonical correlation to investigate the temporal relation between RP, conscious intention, and the dynamics of visual attention to the letter stream used to report intention.

Thus, our experiment aims to shed new light on the relationship between the RP, motor preparation and awareness of intention by overcoming three limitations of many previous studies. First, we use an online probing method that precludes the possibility of postdictive reconstruction of intentions. Second, we avoid the biased sampling problem associated with previous studies of action-locked RPs. Finally, we directly investigate the contribution of visual attention to awareness of intention reports and its relationship to the RP.

Section snippets

Participants

A previous similar study (Schurger et al., 2012) showed a large effect size (dz = 0.9) for the contrast between the pre-stimulus EEG amplitude for responses to a tone that involved lower vs higher reaction times (Schurger A, personal communication). A power analysis for a paired-samples t-test on mean RP amplitudes contrasting the two conditions of main interest in the previous study (see section 2.2. below) indicated a required sample size of 19 participants for a power of β = 0.95 and

Preliminary analysis

Participants executed a self-paced action on average every 11.29 s (SD = 2.53). In the latent awareness task, participants reported an intention to move after 50% of orange probes (SD = 17.97). The average waiting time between self-paced key presses and the frequency of latent awareness reports were negatively correlated (ρ = −0.51, p = 0.02). Participants who pressed more frequently in the self-paced task were also more likely to report awareness of intention in response to an orange letter

Discussion

We performed a novel EEG study of intention to move. Participants in a self-paced movement task were occasionally interrupted by a cue which required them to report whether they were intending to move. An RP-like signal was more conspicuous prior to the interrupting cue when participants reported conscious intention, compared to when they did not. Our study brings new insight about the relations between voluntary action, the RP, and conscious experience. The results broadly support a gradual,

Conclusion

This study provides evidence that the RP is directly related to awareness of intention in voluntary action.

Our findings are neutral regarding the origin of the RP: it may reflect stochastic fluctuations, or it may reflect a signal generated by specific computational operations of voluntary motor planning. However, we found that participants can consciously access the neural processes associated with the RP in advance of action execution, and can then guide their behaviour accordingly. Moreover,

Acknowledgements

This work was supported by European Research Council Advanced Grant (Project Number 323943), HUMVOL, awarded to Patrick Haggard. Patrick Haggard was further supported by a Blaise Pascal International Research Professorship, of the Région Ile-de-France, administered by the Fondation Ecole Normale Supérieure. Elisabeth Parés-Pujolràs was supported by “la Caixa” scholarship, of the “la Caixa” Foundation. Additional support of a joint grant from the John Templeton Foundation and the Fetzer

References (28)

  • A.G. Guggisberg et al.

    Timing and awareness of movement decisions: does consciousness really come too late?

    Front. Hum. Neurosci.

    (2013)
  • P. Haggard

    Human volition: towards a neuroscience of will

    Nat. Rev. Neurosci.

    (2008)
  • M. Kleiner et al.

    What’s new in Psychtoolbox-3?

    Perception

    (2007)
  • H.H. Kornhuber et al.

    Hirnpotentialänderungen bei Willkürbewegungen und passiven Bewegungen des Menschen: Bereitschaftspotential und reafferente Potentiale

    Pflüg. Arch.

    (1965)
  • Cited by (16)

    • Libet's legacy: A primer to the neuroscience of volition

      2024, Neuroscience and Biobehavioral Reviews
    • What is the intention to move and when does it occur?

      2023, Neuroscience and Biobehavioral Reviews
    • A meta-analysis of Libet-style experiments

      2021, Neuroscience and Biobehavioral Reviews
      Citation Excerpt :

      Further methodological alterations yielded temporal patterns that partially differed from the original one (Libet et al., 1983). Some studies used a randomized stream of letters instead of a clock as the method of time measurement (e.g., Haynes, 2011; Parés-Pujolràs et al., 2019). Some studies used functional magnetic resonance imaging (fMRI) instead of an EEG to monitor participants’ brain activity with a higher spatial resolution (e.g., Hirose et al., 2018; Soon et al., 2008), while others only collected behavioral and not neural data (e.g., Giovannelli et al., 2016; Lush et al., 2016).

    • What Is the Readiness Potential?

      2021, Trends in Cognitive Sciences
      Citation Excerpt :

      Subjects were more likely to report a conscious intention when the probe (a sensory cue) was triggered by an RP-like signal, compared with when the probe was triggered by the absence of an RP-like signal. In a similar vein, a different group [39] interrupted subjects at random times while they performed a self-paced motor task. Subjects were instructed to react to the interruption (by pressing a button) only if they happened to be preparing to move at the time.

    View all citing articles on Scopus
    View full text