Complex tools and motor-to-mechanical transformations

Abstract

The ability to use complex tools is thought to depend on multifaceted motor-to-mechanical transformations within the left inferior parietal lobule (IPL), linked to cognitive control over compound actions. Here we show using neuroimaging that demanding transformations of finger movements into proper mechanical movements of functional parts of complex tools invoke significantly the right rather than left rostral IPL, and bilateral posterior-to-mid and left anterior intraparietal sulci. These findings emerged during the functional grasp and tool-use programming phase. The expected engagement of left IPL was partly revealed by traditional region-of-interest analyses, and further modeling/estimations at the hand-independent level. Thus, our results point to a special role of right IPL in supporting sensory-motor spatial mechanisms which enable an effective control of fingers in skillful handling of complex tools. The resulting motor-to-mechanical transformations involve dynamic hand-centered to target-centered reference frame conversions indispensable for efficient interactions with the environment.

Figure 1

Stimuli, apparatus and study design. (a) Examples of stimuli. Two sets of complex, and simple tools in two different sizes, as well as one set of control objects was utilized. (b) Action recipients. We devised three recipient objects to be acted on by all stimulus kinds. (c) Stimuli and action recipients were positioned on the apparatus table by an experimenter before trial onset. (d) Trial structure and timing of tasks in main experiments. We used an event-related design in which different events had variable durations, as depicted within each frame at the bottom of this figure. Participants could view the workspace only in time intervals indicated by the auditory cue. Eyes were always closed during stimulus set up intervals.

Figure 2

Neural correlates of motor-to-mechanical transformations. (a, b) Task-related contrasts revealing significantly different clusters of voxels, with each voxel thresholded at Z > 3.1, and a (corrected) cluster significance threshold of p = 0.05 (controlling for family-wise error rate, FWER), during the transition from grasping to using of (a) complex tools as compared to simple tools, and (b) both simple and complex tools, as compared to control objects, i.e., sticks and twigs, regardless of the hand. Insets with flattened brain surfaces depict significantly active areas in more detail (by the use of parcels from an atlas). Blue dots in insets indicate smoothed borders of the Region of Interest (ROI) shown in (c). (c) ROI analyses in the independently defined tenuicortical supramarginal area (PFt) of the inferior parietal lobule (based on Juelich probabilistic cytoarchitectonic maps, thresholded at 50th% of their maximum probability) and visualized in top row on the right. Middle row: graphical depiction of the results of a significant interaction of phase (Planning, Grasping, Using) and object category (Complex tools, Simple tools, Non-tools) in the 3 × 3 repeated-measures analysis of variance (rmANOVA) for the left PFt (F_{2.447, 46.496} = 5.999, P = 0.0028, ηp² = 0.240, 1-β = 0.906) and right PFt (F_{4, 76} = 6.695, P = 0.0001, ηp² = 0.261, 1-β = 0.990) conducted on neural signals associated with tasks performed with the dominant right hand. Bottom row: this same interaction observed in the left (F_{4, 76} = 1.645, P = 0.172) and right PFt (F_{4, 76} = 4.310, P = 0.0034, ηp² = 0.185, 1-β = 0.915) for tasks performed with the non-dominant left hand. Asterisks indicate significant differences with p-values of ≤ 0.05 (*), 0.01 (**), or 0.001 (***) in post hoc Bonferroni corrected t-tests.

Figure 3

Brain areas involved in planning, grasping and using of complex and simple tools. (a, b) Task-related contrasts revealing significantly different clusters of voxels, thresholded at least at Z > 3.1, and a cluster-corrected (family-wise error rate, FWER controlled) significance threshold of p ≤ 0.05, for (a) planning functional grasps of tools (grasp preparatory processes) contrasted with non-tools, and (b) tool use contrasted with pointing movements with non-tools, regardless of the used hand. (c) ROI (region of interest) analyses in the left caudal middle temporal gyrus (cMTG) and ventral premotor cortex (PMv), with percent signal change (%SC) extracted from 5-mm dimeter spheres centered at coordinates of local signal peaks obtained from an independent tool use localizer (see Methods). A 3 × 3 repeated-measures analysis of variance (rmANOVA) in left cMTG revealed a significant phase (Planning, Grasping, Using) by object category (Complex tools, Simple tools, Non-tools) interaction for both tasks performed with the dominant right hand (F_4,76 = 3.491, P = 0.0113, ηp² = 0.155, 1-β = 0.841) and the non-dominant left hand (F_2.757,52.375 = 4.412, P = 0.0093, ηp² = 0.188, 1-β = 0.827). This same analyses conducted for left PMv revealed neither significant interaction for tasks performed with the right hand (F_2.054,39.023 = 2.877, P = 0.0670) nor the left hand (F_4,76 = 0.608, P = 0.658), although a similar trend was observed only for the dominant hand. Asterisks indicate significant differences with p-values of ≤ 0.05 (*), 0.01 (**), or 0.001 (***) in post hoc Bonferroni corrected t-tests.

Figure 4

Trial structure, timing, and results from the experiment on grasping complex and simple tools outside of the neuroimaging scanner. (a) Four kinds of trial types based on the availability of vision (preview: occluded, non-occluded) and time to the start cue (trial type: delayed, non-delayed) were introduced. For simplicity, they are aligned with the start cue. Tools, either complex or simple, were always presented for at least 0.5 s. Grasping was executed either with no delay (top two rows) or after a 2-s delay (bottom two rows). In trials with non-occluded vision, programming of grasp kinematics, following the start cue, continued with vision available, and was blocked simultaneously with movement onset. Hence, grasping was always performed without visual feedback. In trials with no delay but occluded preview, vision was blocked with the start cue. In delayed and non-occluded trials, following a 2-s interval with no vision available after initial preview, vision was restored with the start cue, and again blocked with movement onset. In delayed and occluded trials, vision was blocked after 0.5-s preview and never restored. Then, grasping triggered by the start cue was performed exclusively based on remembered tool image. We measured movement onsets following to the start cue, and maximum grip aperture (MGA) before target tools were grasped. (b) Results for movement onsets for grasping complex and simple tools. Participants initiated grasping movements significantly faster for complex tools. (c) Grasp kinematics. Top row: MGAs for complex and simple tools contingent on trial type (non-delayed, delayed) and preview (non-occluded, occluded). We found a significant trial type × preview × tool category interaction (see main text for details) wherein MGAs in trials with non-occluded preview and executed with no delay were significantly larger for complex tools, whereas in trials executed following a delay MGAs were significantly larger for simple tools.