The reader can download the full methods (e.g., stimuli, the syntax used to deliver the experiment, the Excel sheets for data processing, the raw and processed data, etc.) used in the current study from the OSF page: https://osf.io/bxcm5/.

Sample

The participants were provided with a link to access the study on the online platform Psytoolkit (https://www.psytoolkit.org/) (Stoet 2010, 2017). Overall, 55 healthy individuals (mean age ± standard deviations: 24.45 years ± 5.18 years; mean education ± standard deviations: 16.67 years ± 3.30 years; 46 females) participated in the study. Participants were recruited through social media (Facebook, Twitter), word of mouth, and personal and professional contacts.

Participants who presented conditions that might influence the reliability of the collected data (e.g. neurological conditions, history of strokes and traumatic brain injury, arthritis, problems with moving hands and feet, visual problems that are not corrected with the use of glasses or lenses, and abuse of any substance that can affect thoughts, moods, and behaviour) were excluded by the study.

Both right-handed and left-handed participants were enrolled because of the absence of an a priori hypothesis on the specific role of right and left-handedness and right and left-footedness. We assessed the handedness and footedness of participants using the self-report questionnaire developed by Coren (1993). We used the data from the Coren’s (1993) self-report questionnaire to define the factor ‘limb’ as dominant or not dominant in the analyses. Our sample was composed of 9 left-handedness and 46 right-handedness participants. Differently, 7 of the participants reported left-footedness, while 48 reported right-footedness.

The interoceptive body awareness, evaluated with the Multidimensional Assessment of Interoceptive Awareness-2 (Mehling et al. 2018), experienced by participants on a group level reflected what is normally observed in the general population (i.e. scores within the 1 standard deviations cut-off). Differently, on an individual level, a few participants showed high interoceptive body awareness for one or a few more scales (i.e. scores beyond the 1 standard deviations cut-off); however, these values were generally just above the cut-off, with most of the scales within the cut-off. Therefore, we can say that all participants experienced normal interoceptive body awareness.

Written informed consent was obtained before participation. The study was designed according to the Declaration of Helsinki and received approval from the local ethical committee at Heriot-Watt University (approval number: 2020–0669-2680).

Instrument usedHand laterality task (HLT) and foot laterality task (FLT)

To explore the more implicit MI abilities of our participants in the upper and lowers limbs, participants were asked to complete the HLT and the FLT, which ask to judge the laterality of hands and feet.

For both HLT and FLT, the conceptualization of the task structure was the same (Fiori et al. 2013; Scarpina et al. 2019). More in detail, right-back/palm and left-back/palm pictures of hands (HLT) and feet (FLT) were presented in four different orientations: 0°; 90°; 180°; 270°. We used these four orientations which allow us to observe two different effects, the stimulus orientation effect and the biomechanical constraints effect (Brusa et al. 2021a, b; Conson et al. 2010; Fiori et al. 2013, 2014; Parsons 1994; Scarpina et al. 2019, 2022a, b). The stimulus orientation effect is derived by right and left limbs at 0° and right and left limbs at 180°. The stimulus orientation effect is an index indicative of visual imagery. While, the biomechanical constraints’ effect, indicating the use of MI, is computed comparing stimuli presented in an awkward position (270° left hand/90° right hand) versus stimuli presented in a comfortable position (90° left hand/270° right hand) (Brusa et al. 2021a, b; Conson et al. 2010; Fiori et al. 2013, 2014; Scarpina et al. 2019, 2022a, b). Overall, 16 pictures per limb (8 pictures of the right limb and 8 pictures of the left limb) in the back or palm perspective are used.

Each task consisted of 96 trials divided into two blocks (48 trials for each block): each stimulus was presented 6 times (3 times in the first block and 3 times in the second block) in a randomised order. To familiarize participants with the task, the two experimental blocks were preceded by one practice block, composed of six stimuli selected randomly from the full data set. Participants seat in front of the computer screen with their left and right index fingers on the “z” and “m” keys of the keyboard. They were asked to judge if the stimulus represented a right or a left limb by pressing, as quickly and as accurately as possible, the “z” key if the picture on the screen was a left limb or the “m” key if the picture was a right hand in one block, and the reverse in the other block (i.e. the “z” key to select right limb or the “m” key to select left limb) to avoid learning effects and ensure constant attention from the participant (Brusa et al. 2021a, b; Scarpina et al. 2019; 2022a, b).

A fixation cross lasting between 1000 and 1500 ms (ms) preceded each trial. The stimulus disappeared as soon as the answer key was pressed by the participant. As previously done, we left a window of 5000 ms, after which the task automatically proceeded to the next trial if the participant did not answer any key (e.g., Curtze et al. 2010). For each trial, RT (correct answers only) and the answers provided by participants were recorded. Average RT in ms and average accuracy (the percentage of correct answers) was calculated for each combination of orientation and posture.

Starting limb, hand or foot, and block order were randomised between subjects.

Mental motor chronometry (MMC) for hands and feet

The more explicit components of MI were explored using the MMC for hands and feet. The MMC task used in the current study was adapted for online administration (Efstathiou et al. 2022). The task is derived from Sirigu et al. (1996) and reflects the version for hands previously adopted for laboratory administration (e.g. Brusa et al. 2021a, b; Scarpina et al. 2019). It comprised two conditions named motor imagery and motor execution. As suggested by the name of the conditions, when it comes to the motor imagery condition, participants were asked to imagine a sequence of movements, differently, during the motor execution condition, participants were asked to execute the same sequence of movements. Participants were asked to imagine and execute the movements with both limbs, hands, and feet. In the case of hands, the movements were: index and thumb opposition; thumb extension from the fist; middle finger crossed on the index finger; and extension of the index and the little fingers. In the case of feet, the movements selected were: foot internal rotation; foot external rotation; foot dorsiflexion, and foot plantar flexion.

The RTs, expressed in ms, needed to imagine and execute each of the four movements were recorded. The RTs collected in the previous studies were recorded using a stopwatch (Schwoebel and Coslett 2005; Sirigu et al. 1996), differently, we used a computerised evaluation that offers a more accurate response (Armitage and Eerola 2020). Similarly to what happened for the laterality tasks, participants had the opportunity to practice the sequence of movements by a video showing each of the movements for each limb and each side. All videos were of the same length (i.e. 2 s), the movement was presented in the third person perspective, and participants could practice the movement as much as they wanted. During each experimental trial, participants were asked to read the instructions, press the spacebar, and with their eyes closed, imagine or execute the movement five times, as accurately and rapidly as possible (Sirigu et al. 1996). Participants were instructed to press the spacebar immediately after the imagery or execution of the target movement. The task was comprised of 32 trials (four movements for each limb, hand, foot, side, left and right, and condition, imagination and execution). The methods of the MMC tasks are available on the OSF page https://osf.io/6kpqx/.

Procedure

Participants had access to the study through a link on the web platform Psytoolkit (Stoet 2010, 2017). On the first screen, participants could read the consent form and give their consent. First, the participants answered a series of screening questions that verified their eligibility for the study. Then, participants were asked to fill out a questionnaire about some demographic variables (e.g., age, gender). After that, participants completed the self-report “A questionnaire to measure hand, foot, eye and ear preference” (Coren 1993). After this first part of the general assessment, participants were invited to take a break (maximum 10 min) before starting the behavioural tasks: the more implicit MI tasks (laterality tasks: HLT and FLT) and the more explicit MI tasks (MMC for hands and feet). The order of group tasks (more implicit versus more explicit) was randomised using two different experimental links. As stated in the “Instrument used” section, participants completed the laterality task for both hands and feet. Limb and block order was randomised between subjects, in this way, we could avoid learning effects and carry-over effects. Between the more implicit and more explicit groups of tasks, a break was allowed (maximum 10 min). Concerning the MMC tasks, as stated in the “instrumented used” section, participants were asked to imagine and execute a series of movements with their eyes closed. More in detail, participants were asked to practice the movements after watching a video of each movement. This was followed by the motor imagery condition for each limb. During these trials, participants were asked to imagine performing the sequence of movements (see “Instrument used” section for target movements), with hands and feet, as quickly and as accurately as possible. After the motor imagery condition, participants completed the motor execution condition for each limb. Differently, in these trials, participants performed the movement, with their eyes closed, as quickly and as accurately as possible. All participants imagined and performed the same movements in the same order (from the index thumb opposition to the extension of the index and little fingers and from foot internal rotation to foot plantarflexion) to avoid cognitive strategies such as counting (Sharma et al. 2009). The order of the limb was randomised. The task was constituted of 32 trials.

Following the behavioural tasks, participants were asked to fill out the Multidimensional Assessment of Interoceptive Awareness-2 questionnaire (Mehling et al. 2018), to obtain a measure of interoceptive body awareness.

At the end of the experiment, a debrief screen appeared, providing participants with useful information about what happened during the study and the contact details of the authors.

The study duration was expected to be 1 h maximum, the required time depended on the duration of the breaks taken by participants, which we left flexible due to the online administration (maximum 10 min).

Analysis pipelinePower analysis

To compute the sample size necessary to observe the effect if present, a power analysis has been performed using G*Power 3.1 (Erdfelder et al. 2009; Faul et al. 2007). We carried out three different power analyses, which were related to our main analyses. The first power analysis carried out (HLT versus FLT) was based on a repeated measure analysis of variance (RM ANOVA). The project was designed as a within-subjects study, with two factors on two levels: Limb (Hand versus Foot) and Posture (Comfortable versus Awkward). The effect size observed for studies using laterality tasks is generally high (e.g. RM ANOVAs ranging from 0.13 to 0.73 η2, see Brusa et al. 2021a, b; Scarpina et al. 2019); however, at the time of the power analysis conducted, there were no available studies with a similar design. Therefore, we opted for a lower effect size (0.20). As for the other parameters, we have selected an a priori power of 1–− β = 0.95, with an alpha error probability of α = 0.05, one group, and four measurements as other input parameters. The output of the present power analysis resulted in a sample size of n = 55, providing a power of 95%.

Differently, for the second power analysis (MMC hands versus MMC feet), the main analysis was based on a correlation. That is because the isochrony component of the MMC tasks (correlation between the time required to imagine and execute the movements) is a key element of the task. The studies from the literature based on the MMC for hands revealed a strong correlation between imagined and executed movements (e.g. RM ANOVAs ranging from 0.53 to 0.70 η2, see Brusa et al. 2021a, b; Scarpina et al. 2019), while no data were available for lower limbs at the time of this power analysis calculation. Then, we opted for an expected correlation of p = 0.50 (correlation p H1). The other parameters adopted were two-tailed, an a priori power of 1–− β = 0.95, with an alpha error probability of α = 0.05, and a correlation p H0 = 0.0. The output of the present power analysis resulted in a sample size of n = 46, providing a power of 95%.

At last, in the third power analysis carried out, we focused our attention on a different aspect of the MMC tasks, the MMC index (Brusa et al. 2021a, b). In this case, the main analysis was an RM ANOVA with a within-subjects study of two factors, each of them on two levels: Limb (Hand versus Foot) and Dominance (Dominant versus Non-Dominant). Since this is the first time a comparison between upper and lower limbs for MMC is made, we opted for a lower effect size (0.20). We selected an a priori power of 1 − β = 0.95, with an alpha error probability of α = 0.05, one group, and four measurements as other input parameters. Similarly to the first power analysis conducted, the sample size required was n = 55, providing a power of 95%.

A sample size of n = 55 participants was chosen. The sample size chosen would ensure that results are reliable and replicable.

Data analysis

Data are stored in OSF: https://osf.io/bxcm5/. Data were analyzed with Statistical Package for Social Science (IBM® SPSS® Statistic, Version 26). The alpha level was set at p < 0.05 for all analyses. For HLT and FLT, for the RTs, trials in which participants gave the wrong response were discarded from the analyses. For the remaining trials, where participants gave the right answer, a cut-off of two standard deviations above and below the individual mean was used to remove outlier responses in other words’ anticipation and/or lack of attention, respectively (Ratcliff 1993; Scarpina et al. 2019).

After removing RTs’ outliers, for accuracy at the HLT and FLT, classic versions, we used a threshold of 50% accuracy for the stimuli displayed at 0° (the easiest stimuli, on which one should not expect errors) to remove responses that could indicate random guessing (Scarpina et al. 2019; Brusa et al. 2021a, b). For the HLT, the average of all four stimuli at 0° was used. In the FLT, we used stimuli displayed at 0° only as this is the most common view for feet. The totality of the responses of the participants was suppressed in such a case.

Accuracy is not a parameter of the MMC tasks; therefore, in the case of the MMC, the RTs of all responses were included; also here, we adopted a cut-off of 2 standard deviations above and below the individual mean to remove those responses indicative of anticipation and/or lack of attention, respectively (Ratcliff 1993; Scarpina et al. 2019).

Four participants were discarded, because they did not survive the data processing. Therefore, new participants were recruited to replace these four through the recruitment to ensure that the sample size of 55 was still met.

More implicit motor imagery tasks (HLT versus FLT)

After data pre-processing, RTs and average accuracy for each orientation (0°; 90°; 180°; 270°) and perspective (palm, back), for the left and right hand separately, have been calculated.

To compare the stimulus orientation effect in hands and feet, we used a 2 by 2 RM ANOVA with Limb (Hand versus Foot) by Angle of rotation (0° versus 180°) as within-subjects factors. The same analysis has been applied to the biomechanical constraints effect, with the only difference in the factor Posture (Comfortable versus Awkward) instead of the factor Angle of rotation (0° versus 180°).

More explicit motor imagery tasks (MMC hands versus MMC feet)

For each participant and each limb, first, we computed the average duration of each movement for the right and the left limb separately, both in the imagery and in the motor execution conditions. From these data, we defined the index for the imagery and motor execution for each side, which we called dominant and non-dominant, and for each limb, hand, and foot. As a first step, we looked for a correlation (i.e. Spearman’s correlation) between imagined and performed actions, for each limb, and separately for the dominant and non-dominant limbs (Brusa et al. 2021a, b; Scarpina et al. 2019). As for previous studies (e.g. Brusa et al. 2021a, b; Scarpina et al. 2019), in the case of a statistically significant correlation, we transformed the correlation coefficient values into z-scores (Steiger 1980). This step allows us to directly compare the correlations’ strength detected. The r-to-z transformation can be performed on the following website http://comparingcorrelations.org/ (Diedenhofen and Musch 2015).

At last, we calculated the overall MMC index (Brusa et al. 2021a, b). To compare hands and feet MMC indexes, we used a 2 by 2 RM ANOVA with Limb (Hand versus Foot) and Dominance (Dominant versus Non-Dominant) as within-subjects factors. Differently from the analyses for the more implicit MI task, here, we considered the factor dominance. In a previous cognate study with the same tasks (HLT and FLT), no differences were observed due to the dominance (Brusa, Erden, Sedda, 2021). Differently, being no studies on limb dominance for the more explicit component of MI, we decided to include here such a factor. We hypothesised a generally better performance for the dominant limb (Curtze et al. 2010; Fiorio et al. 2006; Nico et al. 2004; Takeda et al. 2010).

ResultsMore implicit motor imagery tasks (HLT versus FLT)Stimulus orientation effect

We found a significant main effect of Angle of rotation in RTs (F(1,54) = 125.123, p < 0.001, η2 = 0.70), while Limb (F(1,54) = 0.654, p = 0.422, η2 = 0.01) main effect did not result in significative differences. As expected, participants had faster RTs with stimuli at 0° (Mea n = 1517 ms; ± SE = 53 ms) than with those at 180° (Mea n = 1948 ms; ± SE = 65 ms) independent from the Limb. We found a significant interaction Limb by Angle of rotation (F(1,54) = 7.375, p = 0.009, η2 = 0.12). The interaction, driven by the factor Angle of rotation (Fig. 1), showed faster RTs for both hands stimuli and feet stimuli at 0° (hands: Mea n = 1496 ms; ± SE = 55 ms; feet: Mea n = 1539 ms; ± SE = 66 ms) compared to the ones at 180° (hands: Mea n = 2018 ms; ± SE = 72 ms; feet: Mea n = 1879 ms; ± SE = 80 ms).

Fig. 1

On the top of the figure, we find the graphs comparing the RTs response for stimuli presented at 0° and 180° for hands a and feet b. At the bottom of the figure we find the graphs comparing the accuracy response for stimuli presented at 0° and 180° for hands a) and feet b). The presence of significant differences is represented by lines and asterisks. Bars represent the standard error of the mean. The y-axis represents for the RTs parameter the RTs expressed in ms, and for the accuracy parameter the percentage of correct answers

Regarding the accuracy parameter, as well as for RTs, we observed a significant main effect of Angle of rotation on accuracy (F(1,54) = 16.099, p < 0.001, η2 = 0.23), while Limb (F(1,54) = 2.509, p = 0.119, η2 = 0.04) main effect did not result in significative differences. Participants were more accurate with stimuli at 0° (Mea n = 79.2%; ± SE = 2.3%) than with those at 180° (Mea n = 74.3%; ± SE = 2.4%) independent from the Limb. The interaction Limb by Angle of rotation (F(1,54) = 5.450, p = 0.023, η2 = 0.09) was significant (Fig. 1). The interaction, driven by the factor Angle of rotation, showed greater accuracy of participants in response to hands stimuli shown at 0° (Mea n = 82.2%; ± SE = 2.5%) compared to hands stimuli at 180° (Mea n = 74.3%; ± SE = 2.7%). Differently, the response for stimuli picturing feet was similar between the angle of rotation 0° (Mea n = 76.2%; ± SE = 2.7%) and 180° (Mea n = 74.4% ± SE = 2.5%).

Overall, the results indicate for the RTs parameter the presence of a stable stimulus orientation effect in both hands and feet. Differently, for the accuracy parameter, the typical pattern of more accurate responses for stimuli presented with an angle of rotation of 0° than 180° was observed for hands only.

Biomechanical constraints’ effect

The RTs analysis showed a significant main effect of Posture (F(1,54) = 11.248, p = 0.001, η2 = 0.17): participants were faster with stimuli presented in comfortable positions (Mea n = 1597 ms; ± SE = 61 ms) compared to the ones presented in awkward positions (Mea n = 1698 ms; ± SE = 56 ms), independent from the Limb factor. Differently, the main effect of Limb (F(1,54) = 0.550, p = 0.461, η2 = 0.01) did not result in significative differences. The interaction Limb by Posture (F(1,54) = 16.513, p < 0.001, η2 = 0.23) was significant (Fig. 2). The interaction, driven by the factor Posture, showed the typical pattern of faster RTs for stimuli presented in comfortable positions (Mea n = 1561 ms; ± SE = 66 ms) compared to the ones presented in awkward positions (Mea n = 1781 ms; ± SE = 63 ms) only for hands. The RTs collected for stimuli picturing feet were similar between comfortable (Mea n = 1632 ms; ± SE = 78 ms) and awkward (Mea n = 1615 ms ± SE = 66 ms) positions.

Fig. 2

On the top of the figure, we find the graphs comparing the RTs response for stimuli presented in comfortable positions and at awkward positions for hands a) and feet b). At the bottom of the figure, we find the graphs comparing the accuracy response for stimuli presented in comfortable and awkward positions for hands a) and feet b). The presence of significant differences is represented by lines and asterisks. Bars represent the standard error of the mean. The y-axis represents for the RTs parameter the RTs expressed in ms, and for the accuracy parameter the percentage of correct answers

Similarly to RTs, we found a significant main effect of Posture (F(1,54) = 5.832, p = 0.019, η2 = 0.09), while Limb (F(1,54) = 2.262, p = 0.138, η2 = 0.04) main effect did not result in significative differences. Participants were more accurate with stimuli shown in comfortable positions (Mea n = 77.7%; ± SE = 2.6%) than with those shown in awkward positions (Mea n = 75.2%; ± SE = 2.5%), independent from the Limb. The interaction Limb by Posture (F(1,54) = 4.114, p = 0.047, η2 = 0.07) was significant (