There are two phenomena in motor control that are acknowledged by the whole community: increasing movement speed degrades spatial movement accuracy (Fitts, 1954; Meyer et al., 1988; Schmidt et al., 1979), decreasing movement speed degrades temporal movement accuracy (Newell et al., 1979). However, the exact relation between movement speed and movement accuracy (the speed-accuracy trade-off [SAT]) is still elusive. There are at least five different model accounts of SAT (e.g., Fitts, 1954; Guiard & Rioul, 2015; Hancock & Newell, 1985; Plamondon & Alimi, 1997; Schmidt et al., 1979); none holds for all variations imposed.
Collectively, the development of these accounts helped to provide important insight on the SAT phenomena – each highlighting other models’ limitations. Hancock and Newell (1985) was the only one that encompassed both temporal and spatial error. Furthermore, these authors demonstrated that not only variability (temporal and spatial accuracy) changed as a function of movement speed, but also other moments of the error distribution: constant error (mean), skewness and kurtosis. Guiard and Rioul (2015) account was also the only one that understood the issue that individuals could perform worse than their best. What this means is that, while the outcome will be influenced by the SAT condition, participants are not necessarily performing to optimize the parameters of the given criteria. In more general terms, they provided the first arguments that SAT has the potential to demonstrate large individual bias.
Interestingly, if one complements (and integrates) the aforementioned insights, we find a direction of inquiry that, to the best of our knowledge, was not taken yet. Hancock and Newell (1985) discussed the issue of all moments of error distribution (in space and time) varying as a function of movement speed and posited that this implied a complementarity between space and time (i.e., the complementarity principle). This would mean that to fully understand mean spatial error, for instance, one would need to be aware of all other moments of error in both space and time. This makes sense as time and space are inherently interconnected; a phenomenon that involves speed will influence both space and time. However, when stating their position, they provided functions for each moment of error that would be independent of each other, contradicting the complementarity argument (see Schmidt, 1994). Thus, their operationalization of the complementarity principle failed to encompass the strength of the idea.
Recently, researchers on motor learning and motor control revisited the fact (for the early evidence, see Welford et al., 1969; Woodworth, 1899) that there is individuality in strategies when performing SAT paradigms (e.g., Guiard & Rioul, 2015). Different than Guiard and Rioul (2015), that only considered the fact that individuals might show worse performance than their best in these experiments, these new studies demonstrated that individuals are distinct in their whole tendency to perform, emphasizing either spatial or temporal accuracy (King et al., 2012; Pacheco et al., 2017).
Considering the complementarity principle, these individual preferences (tendencies to favor either space or time in SAT) would affect how the SAT relation is demonstrated in both space and time for a given individual. For instance, if an individual decreases movement time in a task (increasing movement speed) to decrease temporal variance, it would inevitably increase mean temporal error (as there was a change in movement time) and decrease spatial accuracy (as there was an increase in movement speed). Because of distinct individual preferences, changes in task conditions (changes in either spatial or temporal criteria) would affect individuals uniquely. The only way to understand the individual SAT relation would be by also considering the other moments of both spatial and temporal errors. Indeed, this was what Pacheco et al., (2019) demonstrated (see also Hsieh et al., 2015).
Thus, integrating the complementarity principle postulated by Hancock and Newell (1985) and the clear influence of individuality pointed out by Guiard and Rioul (2015), we can postulate that individuals demonstrate preferences to emphasize given moments of distribution of spatial and temporal errors. If this preference accompanies the individual throughout the spectrum of SAT manipulations, then this preference is a strong factor that would predict deviations from a general SAT law – that is still to be unraveled. That is, if an individual consistently emphasizes, for instance, spatial accuracy (variance) over all other moments of errors, then large deviations in spatial bias as well as temporal bias and variance would be observed throughout range of SAT conditions. This would alter how SAT is observed for this individual. Nonetheless, this tendency to emphasize one error moment over others would need to be relatively robust (consistent) to be measured and used as a covariate in understanding SAT trends. To test the robustness of this individual preference is the goal of the present paper. As we are aware, Pacheco et al. (2019) could only provide evidence for the complementarity principle – there was no evidence that the same individual always emphasize a given SAT strategy through all SAT manipulations.
Previous studies demonstrated that individuals can change their way of acting in a given condition if it is required to do so (King et al., 2012; Pacheco et al., 2017). That is, if an individual emphasized temporal accuracy rather than spatial accuracy, one can modify the task constraints to induce changes in how this individual performs the task (i.e., emphasize more spatial accuracy). Nonetheless, there is no evidence that such change in preference for space or time in SAT is permanent. It could be that individuals are able to adapt their strategies for new task requirements transiently, returning to their original preference whenever such requirements cease to be (something that can be called as a shift, see Kostrubiec et al., 2012). In fact, this is a strong possibility. In Pacheco et al. (2017) (see also Hsieh et al., 2013 ), participants had to decrease an error function combining spatial and temporal error in an aiming task. Different conditions weighted more temporal than spatial errors on the performance score (and vice-versa) requiring changes in their preference. While for some individuals the change in preference took time but occurred, others never changed. Additionally, individuals that showed a given preference in one condition were likely to show the same preference in another.
If there are strong individual biases on SAT, we will only understand how speed and accuracy relate to each other after understanding the generalities that emerge from these tendencies. Thus, the present study investigates whether preferences to favor given moments of distribution of spatial and temporal errors are robust signatures of individuals when performing SAT tasks. For this, we followed the same paradigm of previous studies that introduced task constraints that would emphasize spatial and temporal relations different than the individuals’ initial tendency. On the first day of practice, we scanned individual preferences in nine SAT conditions modifying temporal and spatial criteria and, after, asked participants to perform a task that would require changes in their preferences. On the second day, participants performed the same nine SAT conditions to observe whether previous practice modified their preferences.