Neural signatures of voluntary action with long-range intentions

doi:10.21203/rs.3.rs-2083949/v1

Download PDF

Article

Neural signatures of voluntary action with long-range intentions

https://doi.org/10.21203/rs.3.rs-2083949/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this older preprint version

Read the latest preprint version →

Long-range intentions are a vital feature of real-world voluntary action, but have not been extensively studied in relation to their neural correlates. The current study adopted a procedure instead that previous highly repeatable and single decision point paradigms, in which voluntary action (generated by a random ratio (RR), yoked random interval (RI) reinforcement schedule) could be compared with a yoked condition in which participants responded to an external cue. Participants were required to reach the highest reward rates they could in the RI schedule, which offered an indicator of the extent to which long-range intentions have been formed. A classical RP amplitude occurred preceding participants’ keypress action in the current study. EEG amplitudes and EEG variability decreased significantly prior to voluntary action, compared to externally triggered action. These results extend previous findings regarding voluntary action arising from a particular set of long-range intention-based processes, rather than the outcome of stochastic neural fluctuations. Notably, EEG amplitudes decreased significantly differently prior to higher RI-reward rates (i.e., higher plane of long-range intentions formed). The novel experimental paradigm suggests a possible contribution of long-range intentions on the neural activities stage prior to voluntary action.

There are numerous situations in which people, sometimes unknowingly, find themselves producing voluntary actions. Those actions may be driven by long-range intentions, for example, studying hard from the beginning of a semester in preparation for final exams, or/and short-range intentions, such as immediate feelings to act, such as adjusting sitting position (Bratman, 1987; Mulcahy & Call, 2006; Schacter et al., 2007; Pacherie, 2008). In fact, short-range intentions to perform action are highly dependent on the extent to which long-range intentions have formed (Haggard 2008; Pacherie & Haggard, 2010). The relation between long-range intentions and voluntary action has been a pivotal focus regrading ‘conscious intention’ in neuroscience. Conscious intention, in its most fundamental sense, is a change in behaviour brought about by experience. More importantly, both action and conscious intention are driven by common brain activity, namely neural preparation for action, which could explain the correlation between them (Wegner, 2003; Haggard, 2008; Lebreton et al., 2019). Voluntary actions have been investigated as one of the major human action in the field of cognitive neuroscience, with accumulating evidence in support of its distinct functional role (Haggard, 2008; 2018; Passingham et al., 2010).

Probably the earliest reported neural marker of voluntary action is bereitschaftspotential or “readiness potential” (RP). As described by Kornhuber and Deecke (1965), RP refers to a gradual buildup of electrical potential prior to voluntary action. The RP is regarded as “the electro-physiological sign of planning, preparation, and initiation of volitional acts” (Kornhuber & Deecke, 2003); thus, inviting the assumption that the link between the gradual increase in electrical potential and consequent voluntary action reflects a pre-process of establishing intentions to act. Notably, Libet (1982, 1983) attempted to access the neural basis of intention on Wundt’s classical technique, demonstrating that the individual’s neural activity precedes the intent to move by a few hundred milliseconds. Schurger et al. (2012) employed a modified Libet task, in which participants were additionally told to press the button immediately when they heard a brief click as an interruption. Faster responses to these temporally unpredictable cues (the click) are preceded by a gradual negative-going electrical potential over the central area prior to the interruption itself. No similar convergence in slower responses has occurred. The results propose that the neural activities during the RP period reflects a spontaneous subthreshold fluctuation, that would trigger an action onset when it reaches a threshold, suggesting instead that RP is an internal signal of volition in the Libet interpretation (Libet et al., 1983). However, the action in these studies were mainly driven by short-range intentions within a single trial, meaning that the findings may not represent more complicated real-world voluntary action (Churchland, 1981; Khalighinejad et al., 2018; Maoz et al., 2015; Nachev & Hacker, 2014).

Recent studies have adopted paradigms with more specific intentional goals to better resemble voluntary action. Khalighinejad et al. (2018; 2019) adapted a ‘wait-or-skip’ approach previously employed with nonhumans (Murakami et al., 2014), in which the participants could choose to wait or ‘skip’ for big or small rewards. The skip action was regarded as a good example of real-world intentional action, which includes both short-range intentions (the ‘urges’ of being about to act) and relatively long-range intentions about the expected reward of the whole experiment (i.e., the consequence of series actions). Results showed that inter-trial EEG variability decreased more remarkedly prior to voluntary skip action, compared to a control condition when the skip action was produced according to an external cue. These results suggest the reduction in neural noise as a promising neural signature for explaining voluntary action. Moreover, “a consistent process of volition” is detectable during both the early planning stage and the late execution-related stage of action (Khalighinejad et al., 2018; 2019).

A more recent study developed a reinforcement learning paradigm in which the participants learned through trial-and-error about the optimal time to bake a soufflé (Travers et al. 2021). The voluntary action in this study was the key press action to withdraw the soufflé from the oven when it was ready, and the participants were rewarded for the correct timing of their action. This study broadened the research of voluntary action, formed short-range intention (i.e., estimating the baking time) and long-range intention (i.e., action rests on previous action and aims to the final reward) for action at the same time. The results suggested that RP reflects motor planning for upcoming behaviors, rather than freedom from external restrictions.

Although these studies included the factor of long-range intentions, they were not designed to address long-range intentions; therefore, necessary control conditions were lacking to understand the neural signatures specifically associated with long-range intentions. Furthermore, these paradigms have been haunted by traditional stereotyped features, such as having highly repeatable trials, and resting on a single decision point rather than a continuous readout of action, which weakened their ecological validity for representing real-world voluntary action (Miller et al., 2022). To directly address the issue of long-range intentions, we employed a paradigm derived from the schedule of reinforcement studies was adopted (see Ferster & Skinner, 1957). A free-operant reinforcement schedule enables subjects to perform more elaborate voluntary action without any direct trial-based cues.

The current paradigm was adapted from a recently revised version that used random ratio (RR), yoked random interval (RI) reinforcement schedules, for forming intention-based action (Bradshaw & Reed., 2012; Chen et al., 2020; 2021). The RI schedule is presented after the RR schedule, and the reinforcement in the RI block is yoked to the reward on the RR block (see Methods for details). The participants learned to respond at a series of optimal time points on the RI block for the highest points, based on their experiences in the preceding RR block. After the participants were familiar with the RR-yoked-RI schedule, the action released during the RI blocks were voluntary action without any direct single-action-based cues, and namely time-based. Time-based actions have been suggested to be more endogenous than event-based actions, as previous approaches have shown, which significantly reduce endogenous action to externally-triggered ones (Okuda et al., 2007; Pacherie & Haggard, 2010). Thus, actions during the current task are beyond a single trial, and are associated with long-range intentions through a continuous behavioural process: each response in the RI develops from previous response by immediate feedback, is highly impacted by experiences during the RR, and the current contextual information rather than an isolated decision. This procedure allows participants to mentally retrospect previous events and apply the information to the current related events, and the process is namely mental time travel, which is believed as a prime candidate for long-range intentions (Tulving, 1983, 2005; Dudai & Carruthers, 2005; Suddendorf & Corballis, 2007; Pacherie & Haggard, 2010). Thus, the reward rates of participants in RI offer an indicator of the extent to which long-range intentions have been formed. Furthermore, the reinforcement learning paradigm allows behaviour to be modulated by previous experience, having greater flexibility than instinctual modules of behavior, which provides the foundation for reasoning and allows learning in one task to be endogenously transferred to another (Suddendorf & Corballis, 2007; Pacherie & Haggard, 2010).

Based on previous studies on the neural signatures of voluntary action, the present study focused on both the averaged EEG amplitude prior to action onset and the inter-trial variability (Kornhuber & Deecke,1965; Dilek et al., 2022). If the current paradigm is validated, RI-based action (voluntary action) will release a greater plane of conscious intentions than action in a yoked condition with an external-trigger. A difference between EEG variability preceding voluntary action, with variability preceding externally-triggered action at a similar time (see Method), would be predicted, and which would state the crucial neural signatures of voluntary action. Importantly, the possible pattern of EEG amplitudes and/or variability by learning in the present task could further elucidate the potential functional involvement of these signatures, as the nature of the schedule of reinforcement (Tolman & Honzik, 1930; Glimcher, 2005; Gershman, 2019), that is, the comparison between higher and lower level of performance RI blocks (i.e., higher and lower rewards rates blocks) would reveal how these signatures were related to long-range intentions.

In voluntary conditions, participants were asked to press the space bar to earn points on a RR, yoked RI reinforcement schedules. After the participants were familiar with this schedule, the action released during the RI were voluntary action with relatively explicit long-range intentions. Externally-triggered condition came after each RI blocks, participants were instructed to response to an external cue, which is yoked the ones’ responses in the corresponding previous RI-based voluntary action blocks.

Behavioural data

--------------------------

Figure 2 about here

--------------------------

Figure 2(a) shows mean response rates across all experimental blocks for the yoked voluntary condition, as well as these data for each participant. Also shown are the mean numbers of rewards (2b), and reward rates (2c) in each block, along with these data for each participant. All participants became familiar with the task, and developed learned their own action pattern after the first block (reward rates were greater than 0.80 of all possible rewards in the last three bocks). However, the individual data suggest that 9 participants produced performance that varied across blocks, with some reward rates greater than 70%, and also blocks with reward rates less than 70%. These data implied differing performance levels, and may relate to extent of long-range intentions involved in action.

In the “externally triggered” condition, the participants reacted to an external signal which yoked reward responses in the yoked “self -initiated” experimental blocks. Responses made within 1.2s after the signal were identified as “externally-triggered” action ( > = 90% triggered responses for all participants). The average response times of participants were 106 ms (SD = 45.61) in the “voluntary” condition, and 96 ms (SD = 29.73) in the externally-triggered” conditions. The latency to response in the “externally triggered” condition (2.93 s, SD = 1.72) resembled the response time in the yoked “self -initiated” experimental blocks (2.89 s, SD = 0. 95). The SD across trials had a mean of 1.22s and (SD across participants = 0.59) in the voluntary, and 1.42 s (SD = 0.77) in the “externally-triggered” conditions. Based on the above, the yoking design was sufficient (see materials and methods for full details), all ts < 1.

EEG amplitudes analysis

--------------------------

Figure 3 about here

--------------------------

The grand average for the time course and topography of EEG activity prior to a response in the current task reveals a typical negative-going ramp at the central electrodes prior to the onset of action, suggesting classical RP-like amplitudes occurred preceding participants’ action in the current task (Fig. 3). The time course of EEG activity was averaged from central electrodes with prominent negative responses (Cz/1/2/3/4). The topography was calculated within the time window [-0.1 to 0.1], where timepoint 0 indicates action onset.

---------------------------

Figure 4 about here

---------------------------

Mean EEG amplitudes\(\text{f}\)or each condition shows EEG amplitudes were more negative prior to voluntary action (mean change = -0.98 µV, SD = 0.70 µV, p < .001), compared to externally-triggered action (mean change = -0.43 µV, SD = 0.87 µV, p = .068) (Fig. 4a). Cluster-based permutation tests revealed two significantly different clusters of amplitudes between voluntary and externally-triggered action (Cluster 1: -0.65 s to -0.54 s, p < .01; Cluster 2: − .192 s to 0 s, p < .01). Figure 4b shows the change in mean EEG amplitudes between the late (-0.75 s to 0 s) and early (-1.5 s to -0.75 s) time windows for voluntary and externally-triggered action. These data reveal that EEG amplitudes dropped significantly rapidly across this period prior to voluntary action than externally-triggered action, t(15) = 2.81, p = .013. The findings lie in line with accounts of Type I and Type II RPs, which implicated that the possibilities of neural actives prior to voluntary action are more related to long-range intentions-based processes (i.e., planning).

EEG variability analysis

-------------------------------

Figure 5 about here

---------------------------

Measurement and analysis of EEG variability are significant references in studying neural activities prior to action. The decrease of EEG variability is interpreted as the reduction in neural noise in related studies; changes in neural noise level have been linked to decision-making and motivating processes, which are associated with long-range intentions (Kiani et al., 2008; Manohar et al., 2015; Zylberberg et al., 2016; Khalighinejad et al., 2018). Figure 5(a) shows the mean EEG variability across participants for each condition, as represented by the standard deviation of EEG amplitudes, for voluntary and externally-triggered action. Inspection of the shaded grey areas shows that EEG variability prior to the voluntary action was significantly larger than prior to externally-triggered action. Five significant cluster-based permutation tests were identified (Cluster 1: -1.50 s to -0.65 s, p < .01; Cluster 2: -0.50 s to -0.41 s, p = .034; Cluster 3: -0.32 s to -0.27 s, p = .046; Cluster 4: -0.26 s to 0.21 s, p = .047; Cluster 5: -0.08 s to 0 s, p = .033). Figure 5(b) shows the change of average EEG variability between in the late (-0.75 s to 0 s) and early (-1.5 s to -0.75 s) time windows in the voluntary and externally-triggered conditions. EEG variability decreased significantly in both conditions (voluntary mean change = -1.07 µV, SD = .845 µV, p < .01; externally-triggered mean change = -0.502 µV, SD = .392 µV, p < .01). However, the decrease of EEG variability was significantly larger in the voluntary condition than in the externally-triggered condition, t(15) = 3.57, p < .01. Based on the ‘neural noise’ view, these data suggested that voluntary action invoke more viewable functional processes (i.e., long-range intentions-based processes) compared to externally triggered action.

Difference between EEG amplitudes and EEG variability between higher level of performance and lower level of performance blocks

----------------------------

Figure 6 about here

---------------------------

The decrease of EEG amplitude in the two conditions was compared among the 9 participants who showed fluctuating reward rates: higher-level performance (higher reward rates), and lower-level performance (lower reward rates), it may be due to the different strategies that participants conducted, and each strategy point to distinct plane of long-intentions have been formed in actions. Figure 6(a) shows the mean EEG amplitudes for higher- and lower-performance blocks each condition. Figure 6(b) that the decrease of average EEG amplitudes in the late (-0.75 s to 0 s) compared to the early (-1.5 s to -0.75 s) time windows was greater in the higher-level performance blocks than the lower-level performance blocks, t(8) = 2.32, p < .01. Figure 6(c) shows the mean EEG variability across participants for each condition, and Fig. 6(d) shows no difference between higher- and lower performance block between the late and early time, t(8) = 1.41, p = .20.The results revealed that EEG amplitudes dropped more prior to higher-level performance action than lower-level performance action for voluntary action across time windows, and further supported that the level of long-range intention-based have been formed can be reflected in the fluctuation of neural activities.

The current study supports accounts that suggest neural activities period before voluntary action maps onto the cognitive processes of long-range intentions, and differs from EEG patterns observed when action are externally-triggered. Firstly, it was noted that EEG variability and amplitudes decreased to a greater extent prior to voluntary action compared to externally-triggered action. Secondly, there was a trend towards EEG amplitudes decreasing significantly prior to higher-level performance voluntary actions.

EEG amplitudes decreased more prior to voluntary action compared to externally triggered action, hinting at the possible differences in cognitive processes underlying these two actions. We observed the classical pattern of convergence of neural activities prior to movement, replicating previous findings (Kornhuber & Deecke, 1965; Libet et al., 1983; Sinnott-Armstrong & Nadel, 2010; Khalighinejad et al., 2018, 2019). Further, we reported that EEG amplitudes decreased more prior to voluntary action than externally-triggered action over the 600 ms preceding the onset of movement. Accounts of Type I and Type II RPs provide a possible explanation of the current findings (Libert et al., 1983) as reflecting cognitive processes such as planning (believed to be one of the basic features of long-range intention) for voluntary action. Libet et al. (1983) stated that RPs started just ahead of action have lower amplitude (Type I RPs) when participants reported they acted ‘freely and capriciously’. In contrast, Type II RPs started early, compared to Type I RPs, and reached a relatively high amplitude, when participants reported that they had planned before they acted. However, these subjective reports of action associated with Type I and Type II RPs were collected at the end of a block of several trails, and cannot be associated with every single response. However, the current procedure helps expand the analysis away from relying on subjective reports of the types of action performed and make reliance on subject reports unnecessary. Mental time travel theory states that the unique characteristic of human planning, which demonstrates that the reinforcement learning paradigm allows participants to construct the linkage between the previous experiences and future events and the anticipation for future motivational states, that is, long-range intentions had been formed successfully. Therefore, long-range intentions are associated with action occurring during the voluntary action task, where EEG amplitudes were greater compared to prior to externally-triggered action (without explicit long-range intentions). This corroborates speculation that neural signatures prior to voluntary action are more about long-range intention-based processes (Libet et al., 1983).

The ‘neural noise’ view suggests a reduction in neural noise may manifest as a key precursor of voluntary action (Khalighinejad et al., 2018). The important role of active control of internal noise during tasks requiring reaction to external stimuli has been revealed in a series of studies (Kool & Botvinick, 2013; Manohar et al., 2015). Changes in noise level have been linked to multiple cognitive functions associated with long-range intentions, such as decision-making and motivating processes (Kiani et al., 2008; Manohar et al., 2015; Zylberberg et al., 2016). In contrast, if EEG variability is the consequence of stochastic fluctuations in the neural activity, rather than wholly or partly adjusted by the neural noise level, then increasing the contribution of the accumulation process does not produce a greater EEG variability. Since we found the opposite effect, at least in the context of the present task, our results are inconsistent with the explanation of the RP as reflecting a stochastic process for subthreshold action. The current results revealed EEG variability dropped more rapidly in voluntary condition than in externally-triggered condition. The ‘neural noise’ reduction assumption would suggest that, as the neural noise decreased more rapidly prior to voluntary action, involving more evident long-range intentions, compared to externally-triggered action, this further corroborates the value of the reinforcement paradigm in the research of voluntary action. Additionally, the differences between EEG variability prior to voluntary action and externally-triggered action were more obvious during the early time window (-1.5 s to -0.75 s), which may indicate the potential impact of the early stages of neural activities on long-range intention-based action.

It is worth noting that 9/16 participants showed fluctuating reward rates across blocks. According to Gollwitzer (1999), participants in higher-level performance blocks, with higher reward rates, probably behave with a ‘neurotic planner’ strategy, making extensive use of ‘mental time travel’, and ‘imaginatively combining previously stored episodes to generate scenarios concerning action to be performed in the situation’ (see Pacherie & Haggard, 2010). This front-loading strategy retains explicit long-range intentions throughout the entire block, and leaves little room for short-range intentions to occur. In contrast, participants in lower-level performance blocks, with lower reward rates, are more like adopt an ‘optimistic improviser’ strategy, which means participants form long-range intentions over action initiation, without being fully involved in mental time travel, and knowingly form short-range intentions depending on specific circumstances (Pacherie & Haggard, 2010). Thus, the lower level of long-range intentions was formed for participants in lower level of performance blocks.

EEG amplitudes prior to action in higher- and lower-level performance blocks may provide a novel neurophysiological perspective on the intrinsic relations with the long-range intentions and voluntary action. Action in higher-level and lower-level performance blocks (i.e., higher and lower reward rates blocks) are related to the extent to long-range intentions have been formed in measured voluntary action. The current results showed EEG amplitudes decreased more prior to higher-level performance blocks than lower-level performance blocks for voluntary action across time windows. Participants in lower-level performance blocks evolved a lower-plane of long-range intentions (i.e., ‘optimistic improviser’ rather than ‘neurotic planner’ strategies) compared to in higher-level performance blocks, meaning EEG amplitudes prior to action with relatively explicit long-range intentions showed more significant fluctuation.

The reaction time is relatively short due to the limitation of RP in the current study, though we observed significant differences 1.5 s early before the action. However, the reaction time of participants can be easily extended by adjusting parameters for the schedule paradigm (Chen & Reed, 2020, 2021; Chen et al., 2020). The prerequisite to measuring of RP-like amplitudes highlights a fundamental limitation of RP or ERP and related approaches. Identifying RP requires long enough reaction times, otherwise it is almost unobservable. RP allows us to infer changes in neural activities prior to action after prolonged reaction times (e.g., greater than 1.5 s), but not for action with very short reaction time intervals, as action are continuous and coinstantaneous (e.g., typing on a keyboard). This can be avoided by exploring new states of neural activities or advanced recording and analysis approaches.

The current study demonstrated that actions with a greater plane of long-range intentions were preceded by markedly neural amplitudes decrease on RR-yoked RI reinforcement learning paradigm. The result suggests that neural activities prior to actions are related to the extent to which cognitive processes of long-range intentions have been formed, rather than the outcome of stochastic neural fluctuations. The current modified behavioral technique may be useful in extended research linked to conscious intention-based action. Hence, future work could benefit from well-designed behavioral paradigms and the development of rigorous approaches to the phenomenology of human behaviors.

Participants

Twenty-two healthy volunteers were recruited for the present study. All participants were right-handed, had normal or corrected-to-normal vision, and had no history or family history of seizure, epilepsy, or other neurologic or psychiatric disorder by self-report. Six participants were excluded as they did not learn the temporal response pattern after the first RR-RI block, leaving 16 participants (8 female, mean age = 22 years old) for data analysis. The experiment was approved by the local Ethics Committee of Tsinghua University, and each participant’s written informed consent was collected before starting the experiment.

Behavioural task

Participants were tested individually in an electrically shielded chamber, with a 60-inch computer screen (DELL, USA) situation approximately 60 cm in front of the participants. Participants read the study information and paper instructions for the task; the experimental procedure was explained as the EEG cap was being prepared.

The participants then were presented with the behavioural task on the computer. Three stimulus boxes were presented on a white screen (8 cm wide × 3 cm high). The top box (schedule signal) was placed in the centre upper portion of the screen, and displayed a single colour to indicate the operative schedule (RR or RI) throughout each experiment block; either blue or pink, counter-balanced across participants. The middle box was in the center middle portion of the screen, and presented the running points total. The bottom box was in the center lower portion of the screen, and was the response area. Participants were requested to place the mouse cursor in the bottom box, and then press the space bar to earn points.

Participants were told only that different schedules would require different action strategies, but they were not explicitly informed about the type of schedule, and they learned the reinforcement contingency through trial and error. They were given verbal instructions, on the computer screen:

When the task begins, position the cursor over the rectangle which states “click here”, and use the space bar to score as many points as possible. There are eight games in total. The first game is identified with a large blue [pink] rectangle at the top of the screen. When the first game is over, the rectangle will change to pink [blue] to indicate the start of the next game. The rectangles alternate between blue and pink [pink and blue] to indicate the changing games for the remainder of the task. Your goal in each game is to reach the highest score possible. You will see that the points reduce according to the way in which you play but will rise again every so often, according to the pattern of space bar hits that you use. All you need to do is to find the best pattern of space bar hits to score as highly as possible in each game. It may be a good idea to respond quickly sometimes, and slowly at other times, but you need to discover this for yourself!

--------------------------

Figure 1 about here

--------------------------

In the formal experiment, the participants one practice trial of responding on an RR and then a yoked RI schedule to familiarise with the task. They then completed three RR, three yoked RI, and three externally-triggered action blocks (see top panel of Fig. 1). For both the RR-based and RI-based voluntary action blocks, the design was an adaptation of Reed’s RR-RI paradigm (Bradshaw & Reed., 2012; Chen & Reed, 2020; 2021). The RR-based blocks and the RI-based blocks always came in pairs, with RR preceding RI schedule. In the RR schedule, reinforcement was presented for space bar responses according to an RR-10 schedule (i.e., each response had a 1/10 probability of reinforcement). Reinforcement consisted of 40 points being added to the running total shown in the middle box (which started at 10 on each trial). Participants also lost 1 point for each space bar response, regardless of whether the keypress was reinforced. This manipulation was introduced as it has previously been noted that, in the absence of cost for a keypress, there is no penalty for over-responding and, therefore, no reason to regulate performance in line with the contingency of the schedule (see Reed, 2001; Shanks & Dickinson, 1991). In the RI schedule, 40 points were added to the total for a response made after a specified amount of time had elapsed. The RI schedule was yoked to the preceding RR schedule, so that reinforcement in the RI schedule was delivered only after the elapse of the time that it had taken of the corresponding reinforcer to be rewarded on the RR trial. The initial points were set to 10 at the commencement of the task. Each schedule presented was 4 min long. The reinforcement schedule applied in the current study was designed to lead to clear voluntary action without any discrete cues.

In the externally-triggered action blocks that came after each RI block, participants were instructed to respond to an external cue. The cue was an 8 × 3 cm pink-colored box, presented in the centre upper portion of the screen. The box was white when there was no cue. Each presentation of the cue lasted for 16.67ms (i.e., the duration of one frame for a monitor with a 60-Hz refresh rate). Participants were instructed to press the space bar as soon as they detected the pink box flashed. The timing of the cue presentation was yoked to the timing of the participants’ own responses in the corresponding previous RI-based voluntary action blocks. As the participants’ response normally stabilised from the 2nd voluntary action blocks, only three externally-triggered action blocks were added, following the 2nd, 3rd, and 4th RI-based voluntary action blocks. To control for possible confounding effect of memory retention, the participants were asked to complete a 2-min subtraction task (“1567 minus 13” successive computation). There was a 30-sec break between the RI-based voluntary action blocks and the externally-triggered action blocks (Brown et al., 1958; Peterson et al., 1959; Ricker et al., 2016).

Throughout the experiment, participants were required to keep their hands close to the space bar to avoid increasing the reaction time, and to press as quickly as possible. The response hands (left or right) of the participants were counter-balanced across all the participants. After task completion, participants were debriefed and paid.

EEG recording

EEG signals were recorded from 64 electrodes (FPZ, FP1/2,AFz, AF3/4, AF7/8, Fz, F1/2, F3/4, F5/6/, F7/8, FC1/2, FC3/4, FC5/6, FC7/8, Cz, C1/2, C3/4, C5/6, T7/8, CPz, CP1/2, CP3/4, CP5/6, TP7/8, Pz, P1/2, P3/4, P5/6, P7/8, POz, PO3/4, PO5/6, PO7/8, Oz, O1/2), The reference and ground were set to CPz and AFz. A portable wireless EEG amplifier (NueSen. W64, Neuracle, China) was used for data recording at a sampling rate of 250Hz.Electrode impedances were kept below 10 kOhm.

EEG preprocessing

The recorded EEG data were downsampled to 250 Hz and band-pass filtered at 0.1–30 Hz. Then EEG data were re-referenced to the average of the two electrodes near mastoids (T7/T8), following previous studies (Khalighinejad,2018; Travers & Haggard; 2021). The EEG amplitudes trials were defined as the EEG segment starting from − 1.6s before each keypress action to 0.2s afterward, for both voluntary action and externally-triggered action blocks. A [-1.6 s, -1.5 s] baseline was chosen (Khalighinejad et al., 2018; 2019), and the mean amplitudes within the baseline period were subtracted for all the EEG signals per trial. The trials for the rewarded responses in the last three RI-based voluntary action blocks, as well as the trials for all the keypress responses (yoked to the time of the RI responses) in the externally-triggered action blocks.

As the participants might have keypress responses within a relatively short time interval during their exploration, any trials with keypresses within the preceding 1.6s (in other words, two consecutive keypresses within 1.6 s) were rejected. Overall, 19.2\(\pm 14.0\text{\%}\)of the trials were removed for the voluntary action blocks, and 26.9\(\pm 15.0\)% of the trials were removed for the externally-triggered action blocks. Afterward, independent component analysis (ICA) was conducted for artifact rejection. The independent components (ICs) with large weights over frontal or temporal areas, together with a corresponding temporal course showing eye movement or muscle movement activities, were removed. The remaining ICs were then back-projected onto the scalp EEG channels, reconstructing the artifact-free EEG signals. On average, 3–4 ICs were rejected per participant. The segmented trials were subject to an additional artifact rejection procedure: the trials with an absolute value exceeding 100 µV in any of channels of interest. Here, the channels of interest (Cz/1/2/3/4) were chosen covering the centrally-distributed RP-related electrodes (see Wittmann et al., 2014; Goldberg, Busch, & Meer, 2017, and our results). Finally, 185 ± 52.3 (the RI-based voluntary action condition) and 202 ± 54.0 (the externally-triggered action condition) artifact-free trials were obtained.

EEG analysis

The amplitudes of “readiness potential” (denoted as EEG amplitudes) and the corresponding inter-trial standard deviation (denoted as EEG variability, for EEG inter-trial variability) were measured for both voluntary and externally-triggered conditions, and were computed as the key metrics (Khalighinejad et al., 2018). The amplitude was obtained by calculating the median value for each time point across the trial, while the SD was obtained by calculating the standard deviation for each time point across trials. To enhance the signal-to-noise ratio, the metrics were averaged over the channels of interest (Cz/1/2/3/4) (Hinterberger et al., 2013; Pares-Pujoras et al., 2019; Travers et al., 2020). To reveal the EEG signatures for voluntary action, these EEG metrics were compared between the RI-based voluntary action condition and the externally-triggered action condition, by performing a series of paired t-tests for the EEG amplitudes and EEG variability at each sampling point from − 1.5 s to 0 s at the downsampled 250 Hz (the time point zero indicates the onset of action).

A non-parametric cluster-based permutation test was applied to control for multiple comparisons (Maris & Oostenveld, 2007). Specifically, clusters were formed if two or more temporally neighbouring samples showed significant differences (uncorrected p < .05). The cluster-level statistics (cluster-t) was defined as the sum of the t-values within the cluster, and was compared to a permutated distribution, where the conditions of EEG amplitudes were randomly disrupted 20,000 times and the cluster-level statistics were computed as the same statistical procedure as described above. A cluster was considered significant only if the cluster-level statistics were significantly larger than the cluster-level statistics calculated from the permutated data at the 0.05 level. The cluster-based permutation test was performed separately for EEG amplitudes and EEG variability. As accounts of Type I and Type II RPs may offer a potential EEG trajectory across trials, to inspect this distribution, the difference of EEG amplitudes and EEG variability between a late time window (-0.75–0 s) and an early time window (-1.5 – -0.75 s) for each condition was calculated. The change (late time window minus the early time window) was calculated separately for the average EEG amplitudes and EEG variability. The change of EEG amplitudes and EEG variability prior to action was compared between two conditions with a paired t-test.

All participants reached relatively high reward rates (mean reward rates > 0.79), and reached a very high reward rates (> 0.8) in one block after the training block. However, the reward rates fluctuated within an obvious wide range for 9 participants (the difference in reward rates was over 20 percent between blocks after the training block). We thus divided blocks into higher levels of performance (higher reward rates in the RI), and lower level of performance (lower reward rates in the RI). The mean rewards rates for the 9 participants = 0.72; higher level of performance blocks was > 0.72; and lower level of performance blocks was < = 0.72). The comparison between the higher level of performance and lower level of performance RI blocks would reveal neural processes of voluntary action. Then, we calculated the average EEG amplitudes and EEG variability within each block, and averaged EEG amplitudes and EEG variability across blocks within each condition. Finally, the change of EEG amplitudes and EEG variability prior to voluntary action was compared between two rate level conditions with a paired t-test. The calculation process is consistent with the comparison between voluntary action and externally-triggered movements among the selected 9 participants.

All the EEG processing and processing procedures and statistical analyses were manipulated in MATLAB 2017b (MathWorks Inc, Natick, MA, USA) with the help of FieldTrip toolbox (version: 20180101) (Oostenveld et al., 2011). The psychophysics Toolbox extension 3.0 was used to program the task (Brainard & Vision, 1997).

Acknowledgments

This work was supported by China Postdoctoral Science Foundation (2021M701951), Opening Foundation of Key Laboratory of Intelligent Computing & Signal Processing, Ministry of Education, China (2020A005), and the National Natural Science Foundation of China (NSFC) and the German Research Foundation (DFG) in project Crossmodal Learning (NSFC 62061136001/DFG TRR-169/C1).

Bays, P. M. (2014). Noise in neural populations accounts for errors in working memory. Journal of Neuroscience, 34(10), 3632–3645.
Beghetto, R. A., Karwowski, M., & Reiter-Palmon, R. (2020). Intellectual risk taking: A moderating link between creative confidence and creative behavior? Psychology of Aesthetics, Creativity, and the Arts.
Biais, B., Hilton, D., Mazurier, K., & Pouget, S. (2005). Judgemental overconfidence, self-monitoring, and trading performance in an experimental financial market. The Review of economic studies, 72(2), 287–312.
Boland, R. J., & Amonoo, H. L. (2021). Types of Learners. Psychiatric Clinics, 44(2), 141–148.
Bourne, H. (2019). How Confidence Breeds Creativity: A Retired Reporter Breaks Free From Self-Doubt. In Inside Creativity Coaching (pp. 71–75). Routledge.
Bowles, S., Gintis, H., & Osborne, M. (2001). Incentive-enhancing preferences: Personality, behavior, and earnings. American economic review, 91(2), 155–158.
Bradshaw, C. A., & Reed, P. (2012). Relationship between contingency awareness and human performance on random ratio and random interval schedules. Learning and Motivation, 43(1–2), 55–65.
Brainard, D. H., & Vision, S. (1997). The psychophysics toolbox. Spatial vision, 10(4), 433–436.
Bratman, M. (1987). Intention, plans, and practical reason.
Chen, X., Osborne, L. A., & Reed, P. (2020). Role of psychopathic personality traits on the micro-structure of free-operant responding: impacts on goal-directed but not stimulus-drive responses in extinction. Personality and Individual Differences, 163, 110055.
Chen, X., & Reed, P. (2020). Factors controlling the micro-structure of human free-operant behaviour: Bout-initiation and within-bout responses are effected by different aspects of the schedule. Behavioural Processes, 175, 104106.
Chen, X., & Reed, P. (2021). Effect of depression and anxiety on human schedule performance. Learning and Motivation, 75, 101746.
Churchland, P. S. (1981). The timing of sensations: Reply to Libet. Philosophy of Science, 48(3), 492–497.
Cisek, P., & Kalaska, J. F. (2005). Neural correlates of reaching decisions in dorsal premotor cortex: specification of multiple direction choices and final selection of action. Neuron, 45(5), 801–814.
Cohen, J. D., McClure, S. M., & Yu, A. J. (2007). Should I stay or should I go? How the human brain manages the trade-off between exploitation and exploration. Philosophical Transaction of the Royal Society B: Biological Sciences, 362(1481), 933–942.
Colvin, C. R. (1993). " Judgable" people: Personality, behavior, and competing explanations. Journal of Personality and Social Psychology, 64(5), 861.
Daw, N. D., O'doherty, J. P., Dayan, P., Seymour, B., & Dolan, R. J. (2006). Cortical substrates for exploratory decisions in humans. Nature, 441(7095), 876–879.
Deecke, L., & Kornhuber, H. H. (2003). Human freedom, reasoned will, and the brain: The Bereitschaftspotential story. In The Bereitschaftspotential (pp. 283–320). Springer.
Dilek, B., Osumi, M., Nobusako, S., & Morioka, S. (2022). OC10: Effect of painful electrical stimuli on readiness potential in the human brain. Clinical Neurophysiology, 135, e17.
Doyle, M., & Walker, R. (2001). Curved saccade trajectories: Voluntary and reflexive saccades curve away from irrelevant distractors. Experimental brain research, 139(3), 333–344.
Duckworth, A. L., & Steinberg, L. (2015). Unpacking self-control. Child development perspectives, 9(1), 32–37.
Dudai, Y., & Carruthers, M. (2005). Memory: some systems in the brain may be better equipped to handle the future than the past. Nature, 434, 567.
Ferster, C. B., & Skinner, B. F. (1957). Schedules of reinforcement.
Fetsch, C. (2016). The influence of evidence volatility on choice, reaction time and confidence in a perceptual. Elife, 1–31.
Fried, I., Mukamel, R., & Kreiman, G. (2011). Internally generated preactivation of single neurons in human medial frontal cortex predicts volition. Neuron, 69(3), 548–562.
Gershman, S. J. (2019). Uncertainty and exploration. Decision, 6(3), 277.
Glimcher, P. W. (2005). Indeterminacy in brain and behavior. Annual review of psychology, 56(1), 25–56.
Goldberg, M., Busch, N., & van der Meer, E. (2017). The amount of recent action-outcome coupling modulates the mechanisms of the intentional binding effect: A behavioral and ERP study. Consciousness and Cognition, 56, 135–149.
Gollwitzer, P. M. (1999). Implementation intentions: Strong effects of simple plans. American psychologist, 54(7), 493.
Gomes, G. (1999). Volition and the readiness potential. Journal of Consciousness Studies, 6(8–9), 59–76.
Haggard, P. (2008). Human volition: towards a neuroscience of will. Nature Reviews Neuroscience, 9(12), 934–946.
Haynes, J. D. (2011). Decoding and predicting intentions. Annals of the New York Academy of Sciences, 1224(1), 9–21.
Hosking, J. G., Kastman, E. K., Dorfman, H. M., Samanez-Larkin, G. R., Baskin-Sommers, A., Kiehl, K. A., Newman, J. P., & Buckholtz, J. W. (2017). Disrupted prefrontal regulation of striatal subjective value signals in psychopathy. Neuron, 95(1), 221–231. e224.
Human, L. J., Mignault, M.-C., Biesanz, J. C., & Rogers, K. H. (2019). Why are well-adjusted people seen more accurately? The role of personality-behavior congruence in naturalistic social settings. Journal of Personality and Social Psychology, 117(2), 465.
James, W. (1890). The principles of psychology (vol. 1) New York: Henry Holt & Co. Inc. http://dx.doi.org/10.1037/11059-000.
Jo, H.-G., Hinterberger, T., Wittmann, M., Borghardt, T. L., & Schmidt, S. (2013). Spontaneous EEG fluctuations determine the readiness potential: is preconscious brain activation a preparation process to move? Experimental brain research, 231(4), 495–500.
Jo, H.-G., Wittmann, M., Hinterberger, T., & Schmidt, S. (2014). The readiness potential reflects intentional binding. Frontiers in human neuroscience, 8, 421.
Joshi, R. M., Tao, S., Aaron, P., & Quiroz, B. (2012). Cognitive component of componential model of reading applied to different orthographies. Journal of Learning disabilities, 45(5), 480–486.
Khalighinejad, N., Brann, E., Dorgham, A., & Haggard, P. (2019). Dissociating cognitive and motoric precursors of human voluntary action. Journal of cognitive neuroscience, 31(5), 754–767.
Khalighinejad, N., Schurger, A., Desantis, A., Zmigrod, L., & Haggard, P. (2018). Precursor processes of human voluntary action. Neuroimage, 165, 35–47.
Kiani, R., Hanks, T. D., & Shadlen, M. N. (2008). Bounded integration in parietal cortex underlies decisions even when viewing duration is dictated by the environment. Journal of Neuroscience, 28(12), 3017–3029.
Kool, W., & Botvinick, M. (2013). The intrinsic cost of cognitive control. Behav Brain Sci, 36(6), 697–698.
Kornhuber, H. H., & Deecke, L. (1965). Hirnpotentialänderungen bei Willkürbewegungen und passiven Bewegungen des Menschen: Bereitschaftspotential und reafferente Potentiale. Pflüger's Archiv für die gesamte Physiologie des Menschen und der Tiere, 284(1), 1–17.
Lebreton, M., Bacily, K., Palminteri, S., & Engelmann, J. B. (2019). Contextual influence on confidence judgments in human reinforcement learning. PLoS computational biology, 15(4), e1006973.
Libet, B., Gleason, C. A., Wright, E. W., & Pearl, D. K. (1993). Time of conscious intention to act in relation to onset of cerebral activity (readiness-potential). In Neurophysiology of consciousness (pp. 249–268). Springer.
Manohar, S. G., Chong, T. T.-J., Apps, M. A., Batla, A., Stamelou, M., Jarman, P. R., Bhatia, K. P., & Husain, M. (2015). Reward pays the cost of noise reduction in motor and cognitive control. Current Biology, 25(13), 1707–1716.
Maoz, U., Mudrik, L., Rivlin, R., Ross, I., Mamelak, A., & Yaffe, G. (2015). On reporting the onset of the intention to move. Surrounding free will: philosophy, psychology, neuroscience (Mele A, ed), 184–202.
Maris, E., & Oostenveld, R. (2007). Nonparametric statistical testing of EEG-and MEG-data. Journal of neuroscience methods, 164(1), 177–190.
Miller, C. T., Gire, D., Hoke, K., Huk, A. C., Kelley, D., Leopold, D. A., Smear, M. C., Theunissen, F., Yartsev, M., & Niell, C. M. (2022). Natural behavior is the language of the brain. Current Biology, 32(10), R482-R493.
Mulcahy, N. J., & Call, J. (2006). Apes save tools for future use. science, 312(5776), 1038–1040.
Murakami, M., Vicente, M. I., Costa, G. M., & Mainen, Z. F. (2014). Neural antecedents of voluntary action in secondary motor cortex. Nature neuroscience, 17(11), 1574–1582.
Nachev, P., & Hacker, P. (2014). The neural antecedents to voluntary action: a conceptual analysis. Cognitive Neuroscience, 5(3–4), 193–208.
Okuda, J., Fujii, T., Ohtake, H., Tsukiura, T., Yamadori, A., Frith, C. D., & Burgess, P. W. (2007). Differential involvement of regions of rostral prefrontal cortex (Brodmann area 10) in time-and event-based prospective memory. International Journal of Psychophysiology, 64(3), 233–246.
Pacherie, E. (2008). The phenomenology of action: A conceptual framework. Cognition, 107(1), 179–217.
Pacherie, E., & Haggard, P. (2010). What are intentions. Conscious will and responsibility. A tribute to Benjamin Libet, 7, 70–84.
Parés-Pujolràs, E., Kim, Y.-W., Im, C.-H., & Haggard, P. (2019). Latent awareness: early conscious access to motor preparation processes is linked to the readiness potential. Neuroimage, 202, 116140.
Passingham, R. E., Bengtsson, S. L., & Lau, H. C. (2010). Medial frontal cortex: from self-generated action to reflection on one's own performance. Trends in cognitive sciences, 14(1), 16–21.
Saberi-Moghadam, S., Ferrari-Toniolo, S., Ferraina, S., Caminiti, R., & Battaglia-Mayer, A. (2016). Modulation of neural variability in premotor, motor, and posterior parietal cortex during change of motor intention. Journal of Neuroscience, 36(16), 4614–4623.
Schacter, D. L., Addis, D. R., & Buckner, R. L. (2007). Remembering the past to imagine the future: the prospective brain. Nature Reviews Neuroscience, 8(9), 657–661.
Scheinkman, J. A., & Xiong, W. (2003). Overconfidence and speculative bubbles. Journal of political Economy, 111(6), 1183–1220.
Schurger, A., Sarigiannidis, I., Naccache, L., Sitt, J. D., & Dehaene, S. (2015). Cortical activity is more stable when sensory stimuli are consciously perceived. Proceedings of the National Academy of Sciences, 112(16), E2083-E2092.
Schurger, A., Sitt, J. D., & Dehaene, S. (2012). An accumulator model for spontaneous neural activity prior to voluntary movement. Proceedings of the National Academy of Sciences, 109(42), E2904-E2913.
Sinnott-Armstrong, W., & Nadel, L. (2010). Conscious will and responsibility: A tribute to Benjamin Libet. Oxford University Press.
Suddendorf, T., & Corballis, M. C. (2007). The evolution of foresight: What is mental time travel, and is it unique to humans? Behavioral and brain sciences, 30(3), 299–313.
Tolman, E. C., & Honzik, C. H. (1930). Introduction and removal of reward, and maze performance in rats. University of California publications in psychology.
Travers, E., Friedemann, M., & Haggard, P. (2021). The Readiness Potential reflects planning-based expectation, not uncertainty, in the timing of action. Cognitive Neuroscience, 12(1), 14–27.
Travers, E., Khalighinejad, N., Schurger, A., & Haggard, P. (2020). Do readiness potentials happen all the time? Neuroimage, 206, 116286.
Tulving, E. (1983). Elements of episodic memory.
Tulving, E. (2005). Episodic memory and autonoesis: uniquely human?
Wegner, D. M. (2003). The mind's best trick: how we experience conscious will. Trends in cognitive sciences, 7(2), 65–69.
Wittmann, M., Peter, J., Gutina, O., Otten, S., Kohls, N., & Meissner, K. (2014). Individual differences in self-attributed mindfulness levels are related to the experience of time and cognitive self-control. Personality and Individual Differences, 64, 41–45.
Zacharakis, A. L., & Shepherd, D. A. (2001). The nature of information and overconfidence on venture capitalists' decision making. Journal of business venturing, 16(4), 311–332.

There is NO Competing Interest.

Download PDF

Version 1

posted

You are reading this older preprint version

Read the latest preprint version →

Neural signatures of voluntary action with long-range intentions

Status:

Version 1

Abstract

Figures

Introduction

Results

Discussion

Method

Participants

Behavioural task

EEG recording

EEG preprocessing

EEG analysis

Declarations

Acknowledgments

References

Additional Declarations

Status:

Version 1