Libet’s experiment: A complex replication
Introduction
Occasionally, it happens in science that a thought-provoking empirical study becomes substantially influential and stimulates decades of follow-up debates and research. Such is the case of a study published by Benjamin Libet and his colleagues more than 30 years ago (Libet, 1985, Libet et al., 1983, Libet et al., 1982). Their publication initiated heated debates concerning its methodology and interpretations between other researchers (e.g. Libet, 1985, Open Peer Commentary; Dennett and Kinsbourne, 1992, Breitmeyer, 2002, Klein, 2002, Pockett, 2002, Klemm, 2010, Papanicolaou, 2017) and Libet himself responded to many of the discussions in published papers and his 2004 monograph (e.g. Libet, 1985, Author’s Response; Haggard and Libet, 2001, Libet, 2004).
Therefore, it comes as no surprise to learn that some of the technical details of said experiment are, methodologically speaking, not perfect (e.g., see the discussion between Gilberto Gomes and Benjamin Libet in Gomes, 1998, Libet, 2000, Gomes, 2002). Hence, one would expect that there would be a large initiative to replicate Libet’s experiment soon after its publication. Nevertheless, the empirical studies replicating Libet’s experiment in some way or another seem to adopt a different approach. Vast majority of these studies substantially modified or simplified the experimental methodology to either provide support for individual counterarguments against Libet’s conclusions (e.g. Keller and Heckhausen, 1990, Trevena and Miller, 2002, Schurger et al., 2012, Verbaarschot et al., 2015) or expand the study using more advanced technology and procedures (for example using an fMRI machine, see Lau et al., 2004, Lau et al., 2006, Soon et al., 2008).
The more recent the studies are, the more they seem to be focused on specific aspects of the original Libet’s experiment. For example, a study made by Danquah, Farrell, and O’Boyle (2008) aimed at the tactile stimulation which Libet used in his “S series” (see Libet et al., 1982, Libet et al., 1983) and showed that the tactile stimuli may be consciously registered with different latency than visual or auditory stimuli.
Trevena and Miller (2002) presented two experiments focused mainly on analyzing a type of event-related potentials called lateralized readiness potentials (LRPs) instead of the classical readiness potentials (RPs) used in Libet’s case. However, these authors modified the original design and ultimately had to admit that the participants‘ introspective reports of the timing of conscious decision to move differed substantially from what was found by other researchers, including Libet, and that some of these W times were reported after the movement was initiated.
Schurger et al. (2012) introduced a modification, which they called “Libetus interruptus“ design, and showed that the early readiness potentials‘ onsets observed by Libet and his team may be caused by spontaneus fluctuations in neural activity occasionally building up to a movement execution. Additionally, the W reports in their case were similar to what Trevena and Miller (2002) found (and thus different from Libet’s original results).
Other studies focused on biases in reporting the introspective impressions in Libet-style experiments (e.g. Pockett and Miller, 2007, Banks and Isham, 2009, Pockett and Purdy, 2011, Dominik et al., 2017). Some of these showed that certain types of introspective reports (especially the reports of the urge to move) are highly susceptible to being distorted due to changes in the experimental situation.
These and other similar studies are extremely informative for isolated aspects of Libet’s experiment. Nevertheless, their experimental designs are usually notably reduced and do not reflect the complexity of Libet’s original study, which suggests that it might prove useful to conduct a complex replication study. That means a study which does not aim to challenge Libet’s results or interpretations, but instead attempts to conduct the original experiment following Libet’s methodological directions as closely as reasonable.
That is the aim of the present paper. Our procedure consisted of four general steps as follows: (1) we familiarized ourselves with Libet’s original methodological papers (Libet et al., 1982, Libet et al., 1983, Libet, 1985), (2) we devised our own technical plan of the experiment following the original directions, (3) we enhanced the design slightly to overcome some of its original methodological limitations and to adjust it to equipment available to us, and finally (4) conducted the experiment including the data analysis. We aim to publish the research data along this paper to allow other researchers to revise an authentic data sample (see Dominik et al., 2018).
One could argue that our study cannot be called a replication in the strictest sense, since we decided to make some changes to the design. While this might certainly be true, in case of such a complex experiment, it is often difficult to balance replicative accuracy and methodological generalizability, ultimately forcing the researchers to make choices between keeping the design intact, but less valid, and improving it so that it is more valid, but less accurately reproduced. In short, while we are aware that modifying the design may lead to changes in the results, we found some modifications necessary, either for technical or for methodological reasons.
Before reading on, we strongly recommend the reader who is not familiar with the details of Libet’s experiment to read the original papers (Libet et al., 1982, Libet et al., 1983, Libet, 1985). While doing so, we should dedicate a few lines to suggest possible reasons why Libet and his team published the experiment in three separate articles. The answer may be that the authors simply wanted to separate different types of conclusions, as the experiment is notably complex (which is evident from the length of this paper). The first study, published by Libet et al. (1982), emphasized the analysis of the readiness-potentials (RPs). They showed, among other findings, what RPs look like before a spontaneous movement compared to pre-planned movement and that it does not occur before a skin stimulus. The second paper, published by Libet et al. (1983), introduced the introspective reports such as the moment of the first conscious urge to move (called “W” as in “wanting”) or the subjective impression of the actual initiation of the movement (called “M” as in “movement”). The authors pointed out that the RP onsets generally precede not only the movement itself, but also the conscious awareness of wanting to move. The third paper (Libet, 1985) contains mainly discussion and further notes on the results, but it also introduces the concept of a conscious veto and suggests that some series in the original experiment were intended to require participants to deliberately veto an intended movement (Libet, 1985, p. 538).
Section snippets
Participants
Originally, Libet worked with 6 participants (5 of them were females). These were all right-handed college students divided into two groups of three. Libet studied the second group a few months after the study of the first group (Libet et al., 1982, Libet et al., 1983).
Our research sample consisted of 8 participants. This increase in sample size is not large, but the number of 8 participants has a rational reason – it allows a complete rotation of three experimental conditions, which could not
EMG onset timing
The EMG onset is relevant for the W, M and P series. In the W and M series, it constitutes the reference point for the EEG onset and the introspective reports. In the P series, we are interested in the comparison of the target point, the mouse click and the EMG onset.
Discussion
This paper aims to replicate the outcomes of Libet et al., 1982, Libet et al., 1983, Libet, 1985 by conducting a study designed to follow the original methodology as closely as possible. The experiment is complex and as our data suggest, its outcomes are substantially dependent on data collection and analysis procedures. In the following six sections, we discuss technical issues and effects of our methodological choices, as well as the implications of our replication and other recent empirical
Limitations
Present study contains several technical limitations, some of which were already mentioned in the previous sections. Here, we will report additional important technical difficulties which may limit our study.
One limitation is a minor flaw in the mean report calculation from the M and W series in the order (O) mode of recall. The flaw is that not all trials in the O mode of recall contained an EMG onset, even though the button was pressed. The estimate might therefore be slightly inaccurate,
Conclusions
Our study’s goal was to replicate Libet’s experiment and to point out some methodological problems obscured by the experiment’s complexity. To our knowledge, a replication as complex as ours was never carried out, and even though we failed to replicate some elements of Libet’s experiment accurately, we find it critical to present our approach and outcomes. We showed some technical issues in both Libet’s methodology and methodology used by later Libet-style experiments. Our data also showed that
Funding
This research was supported by Palacký University Olomouc, Czech Republic [IGA FF_2017_021].
Acknowledgment
The authors would like to thank Michal Kluka for providing language help and Miroslav Kuba and Jan Kremláček from Faculty of Medicine in Hradec Králové, Charles University for consultations pertaining the ERP analyses.
References (38)
- et al.
Readiness potentials driven by non-motoric processes
Consciousness and Cognition
(2016) In support of Pockett’s critique of Libet’s studies of the time course of consciousness
Consciousness and Cognition
(2002)- et al.
Biases in the subjective timing of perceptual events: Libet et al. (1983) revisited
Consciousness and Cognition
(2008) - et al.
Libet’s experiment: Questioning the validity of measuring the urge to move
Consciouness and Cognition
(2017) The timing of conscious experience: A critical review and reinterpretation of Libet’s research
Consciousness and Cognition
(1998)Problems in the timing of conscious experience
Consciousness and Cognition
(2002)- et al.
Readiness potentials preceding spontaneous motor acts - Voluntary vs involuntary control
Electroencephalography and Clinical Neurophysiology
(1990) Libet’s research on the timing of conscious intention to act: A commentary
Consciousness and Cognition
(2002)Time factors in conscious processes: Reply to Gilberto Gomes
Consciousness and Cognition
(2000)- et al.
Readiness-potentials preceding unrestricted 'spontaneous' vs. pre-planned voluntary acts
Electroencephalography and clinical Neurophysiology
(1982)
On subjective back-referral and how long it takes to become conscious of a stimulus: A reinterpretation of Libet’s data
Consciousness and Cognition
The rotating spot method of timing subjective events
Consciousness and Cognition
Cortical movement preparation before and after a conscious decision to move
Consciousness and Cognition
Lost in time... the search for intentions and readiness potentials
Consciousness and Cognition
We infer rather than perceive the moment we decided to act
Psychological Science
To do or not to do: The neural signature of self-control
The Journal of Neuroscience
“Free will”: Are we all equal? A dynamical perspective of the conscious intention to move
Neuroscience of Consciousness
Time and the observer: The where and when of consciousness in the brain
Behavioral and Brain Sciences
Cited by (9)
Libet's legacy: A primer to the neuroscience of volition
2024, Neuroscience and Biobehavioral ReviewsWhat is the intention to move and when does it occur?
2023, Neuroscience and Biobehavioral ReviewsA meta-analysis of Libet-style experiments
2021, Neuroscience and Biobehavioral ReviewsCitation Excerpt :Second, this approach secures best comparability of the meta-analytical results and those reported by Libet et al. (1983). An EMG recognizes the movement onset approximately 60 ms earlier than an apparatus (Dominik et al., 2018a,b). The weighted mean reported delay in the articles included in the present meta-analysis is 63.09 ms (M = 59.73 ms) and the reported delays range between 40.6 ms (Sirigu et al., 2004) and 76.2 ms (Haggard et al., 2002).
EEG data and introspective reports from the Libet׳s experiment replication
2018, Data in BriefCitation Excerpt :Consecutive sessions were performed about a week apart. The series were organized in the sessions so that the experimental conditions were properly rotated (see Section 2.5, Tables 1 and 2 in our main article [1]). Altogether, seven sessions with each participant took place, while the first session was for training purposes only.
Neurotechnology and Health Law
2023, Casopis Zdravotnickeho Prava a Bioetiky