Abstract
The interplay between attention, alertness, and motor planning is crucial for our manual interactions. To investigate the neural bases of this interaction and challenge the views that attention cannot be disentangled from motor planning, we instructed human volunteers of both sexes to plan and execute reaching movements while attending to the target, while attending elsewhere, or without constraining attention. We recorded reaction times to reach initiation and pupil diameter and interfered with the functions of the medial posterior parietal cortex (mPPC) with online repetitive transcranial magnetic stimulation to test the causal role of this cortical region in the interplay between spatial attention and reaching. We found that mPPC plays a key role in the spatial association of reach planning and covert attention. Moreover, we have found that alertness, measured by pupil size, is a good predictor of the promptness of reach initiation only if we plan a reach to attended targets, and mPPC is causally involved in this coupling. Different from previous understanding, we suggest that mPPC is neither involved in reach planning per se, nor in sustained covert attention in the absence of a reach plan, but it is specifically involved in attention functional to reaching.
Significance Statement
Attention is required to perform dexterous arm movements. In this work, we show the neural bases of the interplay between attention and reaching preparation, with the aim to provide information useful to address effective rehabilitation strategies to treat functional deficits observed in attention-related diseases. We discuss how brain areas are involved in orchestrating attention and reaching by signaling the alignment of their spatial coordinates. Moreover, we found that pupil size changes during reach preparation are related to reach initiation, suggesting a coordination between vigilance and reach promptness when preparing a reach to attended targets.
Introduction
Manual interactions with objects benefit from the concurrent involvement of attention. The posterior parietal cortex (PPC) integrates spatial attention-related information with movement-related one (Fattori et al., 2017; Sulpizio et al., 2023) to perform dexterous reaching movements. In macaques, the medial portion of the PPC includes area V6A (Galletti et al., 1999) which contains cells modulated by covert shifts of attention (Galletti et al., 2010; Caspari et al., 2015) as well as by movement planning (Breveglieri et al., 2016; Fattori et al., 2017). A putative homolog of V6A has been found in the human brain (hV6A; Pitzalis et al., 2013, 2015). Similarly to the monkey V6A, this cortical region is involved in the shifts of covert attention (Vandenberghe et al., 2001; Yantis et al., 2002; Molenberghs et al., 2007; de Haan et al., 2008; Kelley et al., 2008; Capotosto et al., 2013; Ciavarro et al., 2013; Tosoni et al., 2013; Caspari et al., 2018) and in motor planning (Cavina-Pratesi et al., 2010; Vesia et al., 2010; Breveglieri et al., 2021, 2024; Sulpizio et al., 2023).
Given that the hV6A is involved in both spatial attention and movement planning (Galati et al., 2011), and since we commonly move spatial attention when planning a movement unless we are forced to act differently (Rizzolatti et al., 1987), the activation of hV6A during the planning of a movement toward a spatial location could simply be the result of the attentional orienting toward that location. Is this true? Or does hV6A represent the motor plan and the direction of attention, separately? To answer these questions, we examined the role of hV6A in motor planning, either with or without a congruent spatial attention allocation.
It is known that PPC is also involved, besides orienting covert attention, in vigilance and arousal (Galletti et al., 1996; Greene et al., 2014; Lee et al., 2022). Pupil size is a reliable biomarker of the fluctuations of the alerting system, being related to the activity of locus ceruleus (Rajkowski et al., 1994; Aston-Jones and Cohen, 2005; Laeng et al., 2012; Murphy et al., 2014; Reimer et al., 2016; Stitt et al., 2018; van der Wel and van Steenbergen, 2018; Keene et al., 2022; Strauch et al., 2022). Some studies showed that high vigilance states cause pupil dilation (Aston-Jones and Cohen, 2005; Reimer et al., 2016; Mathôt, 2018), and others reported that pupillary light response is enhanced when the bright stimulus is attended versus ignored (Binda et al., 2013; Naber et al., 2013; Binda and Gamlin, 2017; Koevoet et al., 2023a). Moreover, pupil size scales also with motor complexity (van der Wel and van Steenbergen, 2018; Koevoet et al., 2023b). Pupil size has been also used to study the link between arousal and motor actions. For instance, a presaccadic pupil dilation was observed in trials with faster saccadic reaction times (Jainta et al., 2011; Wang et al., 2015, 2016, 2017). It is unknown whether a similar correlation exists between the arousal level, measured by pupil size, and the reaction time of reaching. Neither is it known whether this association is attention-dependent, nor whether hV6A is involved in this process. If this correlation exists, and if it involves hV6A, then an impairment of hV6A should disrupt it.
To check the involvement of hV6A in attention-dependent modifications of pupil size, as well as in devising a motor plan with or without the congruent allocation of attention, we designed a task where motor planning and attention can be manipulated independently of each other. We used transcranial magnetic stimulation (TMS) to establish a causal role for the hV6A. We found that hV6A is not involved in motor initiation per se, but specifically when attention is endogenously allocated to the reaching target during planning. Furthermore, we show here a correlation between the level of arousal and reaching initiation and a causal role of hV6A in this coupling, only in the case of reaching attended locations.
Materials and Methods
Participants
Thirty-four healthy volunteers participated in this study. Seventeen of them took part in a transcranial magnetic stimulation (TMS) experiment (age, 23.12 ± 3.37 years; age range, 19–30 years; 5 males), whereas the remaining 17 (age, 27.82 ± 7.05; age range, 23–48 years; 7 males) participated in a control experiment. The participants were classified as right-handed based on the Edinburgh Handedness Inventory (Oldfield, 1971), had normal or corrected-to-normal visual acuity in both eyes, and were naive as to the purposes of the experiment. None of the participants had neurological, psychiatric, or other medical problems, nor did the participants of the TMS experiment have any contraindications to TMS (Rossi et al., 2009). Participants provided written informed consent. The procedures were approved by the Bioethical Committee at the University of Bologna (Prot. 170133, Prot. 237243, Prot. 57635) and were in accordance with the ethical standards of the 2013 Declaration of Helsinki. No discomfort or adverse effects during TMS were reported or noticed.
TMS experiment: localization of brain sites
The coil position was identified on each participant's scalp using the Cortexplore Neuronavigator (Cortexplore; Klink et al., 2021; Breveglieri et al., 2024).
We tested two active stimulation sites, the area of interest (left hV6A) and a control area (V1/V2), and one Sham condition. The Talairach coordinates for hV6A we used were x = −10, y = −78, and z = 40 (Talairach and Tournoux, 1988; Ciavarro et al., 2013; Breveglieri et al., 2021, 2023a,b, 2024), which were similar to those used for studying the anterior part of the superior parieto-occipital cortex (Vesia et al., 2010, 2017), a region that likely includes hV6A (Pitzalis et al., 2015) and that was investigated in several imaging studies (Filimon et al., 2009; Cavina-Pratesi et al., 2010; Gallivan et al., 2011; Tosoni et al., 2015). To target V1/V2, the coil was centered 2 cm above the center of the inion, thus resulting in a bilateral stimulation (Romei et al., 2016; Chiappini et al., 2018). Sham stimulation was performed by placing the coil tilted at 90° over the vertex bilaterally, so that participants could feel coil–scalp contact and discharge noise as during active stimulation, but no current was induced in the brain (Lisanby et al., 2001; Sandrini et al., 2011). Bilateral control conditions are often performed (Vesia et al., 2010; Breveglieri et al., 2021, 2023b, 2024).
TMS protocol
Biphasic TMS pulses (10 Hz, three pulses, as performed in other studies on the medial PPC; Vesia et al., 2010; Striemer et al., 2011; Breveglieri et al., 2024) were delivered using a Deymed DuoMAG XT stimulator connected to a 70 mm figure-of-eight coil. Stimulation of hV6A was carried out by placing the coil tangentially over the scalp site along a parasagittal line with the handle pointing downward (Vesia et al., 2010; Breveglieri et al., 2021, 2023a,b, 2024). The active control area (V1/V2) was targeted by placing the coil tangentially over the scalp site along a parasagittal line with the handle pointing downward.
To set TMS intensity, the resting motor threshold (rMT) was estimated for all participants in a preliminary phase of the experiment using standard procedures (Sandrini et al., 2011). Motor-evoked potentials (MEPs) induced by stimulation of the left motor cortex were recorded from the right first dorsal interosseous muscle (FDI) by means of a two-channel DuoMAG MEP amplifier. Electromyography signals were finite impulse response (FIR)-filtered and digitized at a sampling rate of 5 kHz. Pairs of disposable pre-gelled Ag–AgCl surface electrodes were placed in a belly tendon montage with a ground electrode on the midpoint of the palmar surface of the wrist. The optimal scalp position for inducing MEPs from the right FDI was first localized, and the rMT was determined from that position. The rMT was defined as the minimal intensity of stimulator output that produced MEPs with an amplitude of at least 50 μV in the FDI with a probability of 50% (Rossini et al., 2015). For both hV6A and V1/V2 stimulations, the intensity of magnetic stimulation was fixed at 120% of the rMT, as in a previous study (Breveglieri et al., 2023b). The range of intensities was 50–71% of the total stimulator output (mean value, 61.53 ± 6.03). No phosphenes were perceived by the participants.
TMS experiment: apparatus and behavioral task
We used a setup that consisted of a 19-inch touchscreen (37.5 cm × 30 cm, ELO IntelliTouch 1939 L, 1,280 × 1,024 pixel screen resolution) set vertically at 43 cm in front of the participants on a desk. The screen displayed the targets of the reaching movements performed by the participants (Fig. 1A, gray squares). For stimuli presentation, we used Matlab (MathWorks RRID:SCR_001622) with the Psychophysics toolbox extension (Brainard, 1997). Participants were seated on a comfortable chair in a darkened room, with their heads stabilized by a head/chin rest to minimize head movements. In all trials, the reaching movement started with the participant's right hand on a button [home button (HB); Fig. 1A] placed on the desk, centered on the touchscreen (Fig. 1A).
A, Timeline of attention/reaching task. Fix, fixation time; cue, cue onset; plan, delay between cue onset and go signal; go, go signal (a small vertical line), Reaction time and reaching. The cue is depicted larger than the targets for the reader's convenience (real dimensions are stated in the Materials and Methods section) and colored in orange and blue (color-blinded people's convenience, real colors are stated in the Materials and Methods). The timeline is shown for MotorATN congruent trials (top), in which attention and motor plan were directed toward the same hemifield; for MotorATN incongruent trials (middle), in which attention and motor plan were directed toward opposite hemifields; and for Motor valid trials (bottom), in which attention was not constrained, the direction of the motor plan was instructed, and the go signal appeared in the same hemifield as the motor plan. The same timeline also applied to Motor invalid trials (not shown for conciseness). B, Types of trials, according to the information received by the central cue: MotorATN trials and Motor trials. The MotorATN trials were only valid (go signal in the target where attention was directed by the colored side of the cue) and could be congruent (attention and movement plan directed toward the same location) or incongruent (attention and movement plan aimed in opposite directions). The Motor trials could be valid (go signal in the target where the movement was planned) or invalid (go signal in the opposite target).
The task was designed to associate or separate the direction of spatial attention from the direction of the movement plan and was adapted from a task first used in monkeys (Messinger et al., 2021). To this aim, we used a cue whose color instructed participants about the direction of the movement to plan and whose side about the deployment of attention. Each trial started, after an intertrial period of 6 s, with the onset of the fixation point (diameter, 0.3 cm, 0.4° of visual angle) in the center of the screen between the two reaching targets. This indicated to the participants to press and hold down the home button. The two reaching targets (squares of 0.6 cm side, 0.78°, located 10° lateral to the fixation point) were displayed on the touchscreen during the entire duration of each trial. After a fixation period (Fig. 1A, fix) of 1.3–1.5 s (randomly chosen) a central, endogenous cue (0.6 cm side, 0.76°) appeared around the fixation point, which informed the participant about the target to covertly attend and the target to subsequently reach [Fig. 1B, motor–attention (MotorATN) trials, where the direction of attention and of the motor plan was constrained] or only about the target to subsequently reach [Fig. 1B, motor (Motor) trials, where the direction of the motor plan was instructed whereas attention wasn't]. After a randomly chosen period of 0.3–0.6–1 s (plan), a small vertical line (go) appeared for 0.08 s in the center of one reaching the target. Importantly, the two trials with the duration of the epoch plan of 1 s have been inserted (and then excluded from the analysis) in each condition/stimulation area to guarantee the participants’ attention to the cued side. In fact, the more variable the duration of this epoch, the less the probability of time-locked behaviors of the participants. TMS pulses were delivered during the plan epoch, with the first pulse delivered after 50 ms from the cue onset. After a variable reaction time to the detection of the go signal, participants reached with their right hand the previously cued target (Fig. 1A, cued by the color, reaching). At the movement offset, the targets and the fixation point disappeared and another intertrial period started.
In MotorATN trials (Fig. 1B, left), the color of the cue was informative about the location of the target to subsequently reach, in order to make participants plan a reach toward the cued side (red, reach planning to the right target, represented in orange in Fig. 1; green, reach planning to the left target, represented in blue in Fig. 1), while the colored side of the cue was informative about the location of the subsequent onset of the go signal, in order to make participants covertly move the spatial attention toward the cued side (right side colored, the go signal will appear within the right target; left side colored, the go signal will appear within the left target). In MotorATN trials, the go signal appeared always within the attended target (instructed by the colored side of the cue, Fig. 1A), so all the trials were “valid” as in the Posner paradigm (Posner, 1980). If the movement was planned in the same location (as instructed by the cue color), the MotorATN trial was labeled “congruent” (Fig. 1A, top, and Fig. 1B, left); if the movement was planned in the opposite, unattended target, the MotorATN trial was labeled “incongruent” (Fig. 1A, bottom, and Fig. 1B, left). Thus, in congruent trials, the participants had to plan a reach toward an attended location during the plan epoch, whereas in incongruent trials, they planned a reach toward an unattended location. In MotorATN trials, the attention was constrained to one side of the screen.
In Motor trials (Fig. 1A, bottom, and Fig. 1B, right), the central cue was a fully colored square informative only about the location of the movement plan (same color conventions as in MotorATN trials). In these trials, participants had to plan a reach without any constraints concerning the location where attention must be directed during the plan epoch. Effectively, in these trials, the cue was neutral regarding attention. To ensure that the attention of the participants was not automatically directed to the location of the movement plan, we designed and inserted valid and invalid Motor trials in equal numbers. In valid trials, the go signal appeared in the target of the planned movement (Fig. 1A, bottom). Conversely, in invalid trials, the go signal appeared in the opposite target (not shown in the figure). Overall, eight conditions were tested (four conditions with MotorATN trials and four for Motor trials, Fig. 1B). Importantly, attention was not constrained during the movement execution, either in MotorATN trials or in Motor ones.
The task was composed of two blocks of 48 trials each (6 trials per condition per block) per stimulation condition (Sham, V1/V2, hV6A, counterbalanced), for a total of 288 trials performed over the same experimental session. Each session lasted approximately 2 h. The task was always performed with the right arm. We randomized the conditions of MotorATN trials and of Motor trials (they were interleaved in each block). A 48-trial training block was included at the beginning of the experimental session.
TMS experiment: data acquisition, analysis, and statistics
The kinematics of reaching movements was recorded using a motion tracking system (VICON motion capture system, 5 M cameras, 1,024 × 1,024 pixel resolution) by sampling the position of two markers at a frequency of 100 Hz; markers were attached to the right wrist (on the scaphoid bone) and to the nail of the right index finger (reaching finger). Reaching onset/offset was determined as the time when the markers’ velocities exceeded/fell and remained below 30 mm/s. The reaction time was defined as the interval between the “go” signal offset and reaching onset. Participants were asked to move their hands without pauses or interruptions, at a fast but comfortable speed, and as accurately as possible.
Given the intrinsic difficulty of the task, a possibility existed that participants reached the wrong target or started a wrong movement trajectory and amended it to get to the correct target. We excluded a trial from the subsequent analyses if the endpoints were on the opposite side of the cued target and if the first or the second half of the trajectories exceeded the two standard deviations calculated with all the trajectories of that participant. We also excluded trials with RTs shorter than 100 ms (Ciavarro et al., 2013) or longer than 1,000 ms (Rizzolatti et al., 1987). We excluded ∼6% of trials for at least one of these abovementioned reasons. The 22% of excluded trials were congruent trials, the 28% incongruent trials, the 24% unconstrained trials, and the 32% invalid ones.
We used an eye tracker (EyeLink 1000, SR Research) to record real-time gaze position and pupil size at 1 kHz. Before collecting data from each participant, the equipment was calibrated using a nine-point grid (horizontal distance, 8 cm; vertical distance, 5 cm) that the participants were asked to fixate steadily (3 × 3° tolerance window) and to covertly attend the targets.
TMS experiment: analysis of behavioral variables
The influence of the stimulation on reaction times in the different trial types was evaluated separately in valid and invalid trials, because in valid trials, no redirection of attention to different hemifields occurred, whereas it was the case in invalid trials at the appearance of the go signal in the opposite hemifield than the one where participants were planning a movement.
In valid trials, we used a two-way repeated measures analysis of variance (ANOVA) with TMS (three levels: Sham, V1/V2, and hV6A) and trial type (three levels: MotorATN congruent, MotorATN incongruent, and Motor valid trials) as factors. In invalid trials, we performed a two-way repeated measures ANOVA with TMS (three levels: Sham, V1/V2, and hV6A) and redirection side (two levels: rightward and leftward) as factors.
In all the analyses, the threshold for significance was set at 0.05, and all post hocs were carried out with the Duncan correction for multiple comparisons.
TMS experiment: analysis of pupil size
As the pupil size is considered an index of effort and attention (Morad et al., 2000; Paladini et al., 2016; Keene et al., 2022), we have tested the changes in pupil size during the plan epoch. Following the procedures of baseline correction used previously (Bala and Takahashi, 2000; Moresi et al., 2008; Cherng et al., 2020; Hsu et al., 2021) for each trial, a baseline value was determined by averaging the pupil size of the last 100 ms before the cue onset. To rule out the influence of the color of the cue on pupil size, we averaged the pupil size of the two congruent conditions, of the two incongruent ones, and of the conditions where attention was not constrained. Data were not normally distributed (Shapiro–Wilk test, p < 0.05), so we used a nonparametric ANOVA [SPM1d Matlab package (Pataky, 2012), codes at www.spm1d.org] with factor TMS (three levels: Sham, V1/V2, and hV6A), to compare the pupil size during the plan epoch of the different stimulation conditions in the different trial types.
To investigate whether the pupil response was predictive of the reaction time, we have used linear mixed-effects (LME) models as performed previously (Koevoet et al., 2023a) in each trial type. To account for interindividual differences in pupil size and to isolate evoked pupil response from baseline pupil size, we robustly z-scored the pupil size (Rousseeuw and Hubert, 2011; Koevoet et al., 2023a) by subtracting the median baseline pupil size from the data of the last 100 ms before the go signal and subsequently dividing by the median absolute deviation per participant. We then included in the model the interaction between pupil response and TMS with a Matlab formula: reaction time ∼1 + pupil response*TMS + (1 | participant). Next, in case the interaction term was significant (see Results), we ran the model during Sham stimulation [formula: reaction time ∼1 + pupil response + (1 | participant)]. If this correlation was significant (p < 0.05), we tested if it was still significant during V1/V2 or hV6A stimulation.
Control experiment
In the control experiment, we collected the pupil size of participants using the same apparatus and visual stimuli used in the TMS experiment. The task sequence was the same as in the TMS experiment except for the timing of the cue appearance and for the events after the cue offset. In this experiment, the cue appeared for 1 s, and participants were instructed to detect its offset by releasing the home button. Six conditions were tested (10 trials each), four of them with the same half-colored cues of the MotorATN trials of the TMS experiment and two with the full-colored cue of the Motor trials of the TMS experiment. Nevertheless, in this experiment, the color and the shape of the cue were neither informative about any attentional directing nor about any spatial motor plan, and the participants did not take part in the TMS experiment. This control experiment was conceived to see whether there are differences in pupil size dynamics for visual stimulation which instructed, or did not instruct, the direction of covert attention. Differences in pupil size dynamics between the control experiment and the Sham condition of the TMS experiment were tested via nonparametric ANOVA [SPM1d Matlab package (Pataky, 2012)] with factor experiment (two levels: TMS or control) in each trial type.
Results
Effectiveness of the cue in directing attention
The double informative nature of the cue of our task had already been revealed effective in nonhuman primates (Messinger et al., 2021) in instructing attention and/or motor plans. However, as our task was a simplification of the task of Messinger et al. (2021), we wanted to confirm whether the attention of participants was directed as instructed during the plan epoch. To do this, we used the reaction time as an indirect index of the direction of attention, as in the classic test of Posner (Posner, 1980).
Thus, we measured the reaction times of participants to the detection of go signal (reach initiation) in the different trial types of TMS experiment during Sham stimulation (Fig. 2A). Reaction times turned out to be affected by trial type (one-way ANOVA, F(3,48) = 12.30, partial η2 = 0.43, p < 0.001) in that reaction times of congruent trials were significantly faster than the ones of all other trial types (all p < 0.02). This confirmed the expectations that the common location of spatial attention and motor plan represents a gain that improves the detection of the go signal. Participants were also slower in detection in invalid trials than in congruent and incongruent trials (all p < 0.01), again as expected. These results confirm the effectiveness of our attentional manipulation by demonstrating increased detection during congruent compared with incongruent, unconstrained, and invalid trials. We thus confirm that the cue features (side and color) were effective in directing the attention of participants as instructed in our task design.
A, Reaction times during Sham stimulation in the different types of trials of the TMS experiment. The bars represent standard error, and the asterisks represent significant (p < 0.05) post hoc comparisons. The gray lines connect points that represent the data of individual participants. These data show that the task elicited attention in the expected way. B, Reaction times of different types of valid trials in the different stimulation conditions (Sham, black; V1/V2, white; hV6A, gray). It is evident the effect of the stimulation in slowing down the detection of the go signal for reaching. This figure contains only valid trials (MotorATN congruent, MotorATN incongruent, and Motor unconstrained trials). Individual participants’ data are in Figure 3. Data regarding Motor invalid trials are shown in Figures 6 and 7.
Valid trials
hV6A stimulation affected reach initiation in congruent motor–attention trials
As shown in Figure 2B, the stimulation of hV6A and V1/V2 produced significant effects on reaction times in valid trials (interaction TMS by type of trial, F(4,64) = 2.98, partial η2 = 0.16, p = 0.03, individual participants’ data in Fig. 3). The gain in reaction time brought by the colocalization of the motor plan and of attention seen during Sham stimulation (Fig. 2B, black columns, all p < 0.01) was canceled by the stimulation of either V1/V2 or of hV6A (Fig. 2B, white and gray columns). After V1/V2 or hV6A stimulations, all the reaction times were similar and did not depend on the trial type (all p > 0.20). Moreover, in congruent trials, the reaction times during Sham stimulation were different from those after hV6A (p = 0.04) or V1/V2 stimulation (p = 0.04), whereas reaction times after hV6A or after V1/V2 stimulation were not different (p = 0.85). In all the other trial types (incongruent attention-reach plan and reaching with unconstrained attention), the stimulation did not affect reaction times (all p > 0.05), suggesting that neither hV6A nor V1/V2 were causally involved in reach initiation when the movement was planned in an unattended location or when attention was not constrained to the target (Figs. 2C, 3).
Mean population reaction times with data of individual participants in valid trials. The same conventions as in Figure 2.
In summary, either stimulation of hV6A or of V1/V2 led to an increase in reaction times to the detection of a visual peripheral target (go signal), specifically when attention and motor plan were on the same side. This suggests that these areas are causally involved in sending information to the motor cortex about the alignment of the spatial coordinates of attention and motor plans.
hV6A stimulation did not affect pupil size
We wanted to test whether the stimulation of hV6A affected the arousal level, measured through pupil size. Pupil size was not affected by the stimulation (nonparametric repeated measures ANOVA with factor TMS, p > 0.05), and this was true in all the trial types (Fig. 4). This suggests that neither hV6A nor V1/V2 is causally involved in modulating pupil size per se.
Pupil size dynamics during the plan epoch of the different trial types of valid trials in the TMS experiment and in the control experiment. Left, Pupil size is represented as baseline corrected values (see Materials and Methods). Different colors represent different stimulation conditions, and the yellow trace represents pupil size dynamics during the conditions of the control experiment with the cue of the same features of the corresponding TMS trial. The black thick lines represent the time when the pupil size of control trials was significantly different from the one during Sham stimulation. Right, Differential values between pupil size during each stimulation condition and pupil size of the control experiment are plotted over time. Pupil constriction is evident in all the trials, but it is more intense during the control trials, when participants paid attention to the cue. No effect of stimulation was found.
Pupil size changed during the plan epoch (Fig. 4) with a pupil constriction due to the pupillary light response, with a well-known time course (Wang et al., 2015). The comparison of the pupil size during the Sham stimulation with the pupil size of a group of participants of a control experiment (where participants looked at the cues that conveyed the same illumination as in the TMS experiment, but neither being informative about the attentional orienting nor instructing a reach planning) revealed that the pupil size of participants looking at the uninformative cue (control experiment; Fig. 4, left, yellow traces) was significantly lower than during the observation of the informative cue during Sham stimulation (TMS experiment; Fig. 4, left, black traces; the differential values are plotted in Fig. 4, right). Moreover, in the TMS experiment, before pupil contraction (which occurred after 300 ms from the cue onset), a slight pupil enlargement was observed, which was significant from ∼220 ms after the cue onset in congruent and incongruent trials and even before in unconstrained trials (p < 0.05; Fig. 4, left). As larger pupil sizes have been associated with orienting responses (Wang et al., 2015; Hsu et al., 2021), this suggests that the task used in the TMS experiment was very effective in orienting the attention of the participants.
hV6A stimulation affected coordination between arousal and reaching in congruent motor–attention trials
To test whether a correlation arousal-reach initiation does exist and, in this case, the role of hV6A in this coupling, we performed a trial-by-trial correlational analysis using linear mixed-effects models (LME) between pupil response right before the go signal and reaction time. In congruent MotorATN trials, the LME model showed a significant pupil response main effect (β = 12.87 ± 5.73, t = 2.24, p = 0.03) and a significant interaction pupil response by TMS (β = −6.65 ± 2.65, t = −2.51, p = 0.01). Thus, we ran the model for each TMS condition to evaluate whether pupil size was a good predictor of reaction time. During Sham stimulation, pupil response significantly predicted the reaction time (β = 10.41 ± 3.36, t = 3.10, p = 0.002, Fig. 5A), in that larger constrictions (lower z-scored values) led to faster reaction times. This was expected because larger constrictions indicate stronger attentional orienting (Naber et al., 2013; Binda and Gamlin, 2017; Koevoet et al., 2023a) and stronger attentional orienting causes faster reaction times. The stimulation affected this correlation, but with different effects depending on the stimulated area. After V1/V2 stimulation, the significant correlation remained, but with an opposite trend (β = −12.64 ± 5.01, t = −2.52, p = 0.01, Fig. 5B) in that larger dilation led to faster reaction times. Instead, after hV6A stimulation, the correlation between pupil size and performance was no more significant (β = −3.26 ± 4.06, t = −0.80, p = 0.42, Fig. 5C). In incongruent MotorATN and unconstrained (Motor) trials, pupil size did not significantly predict the reaction time in any stimulation condition, because neither the main effect of pupil response nor the interaction pupil size by TMS was significant (incongruent trials: main effect of pupil response, β = 4.12 ± 5.00, t = 0.84, p = 0.40; interaction pupil response by TMS, β = −2.09 ± 2.27, t = −0.92, p = 0.36. Unconstrained trials: main effect of pupil response, β = 0.50 ± 4.91, t = 0.10, p = 0.92; interaction pupil response by TMS, β = −0.71 ± 2.22, t = −0.32, p = 0.74).
Pupil responses predict reaching reaction times in congruent trials. Significant prediction after Sham stimulation (A), after V1/V2 stimulation (B), and nonsignificant prediction after hV6A stimulation (C). The light gray lines are linear regression fits to data per participant. The thick lines show the correlations pooled over all trials. *p ≤ 0.01; n.s., p > 0.05.
To summarize, there is a coordination mechanism between arousal level (indicated by the pupil size) and reaction time of reaching toward attended locations, where greater pupil constrictions predict faster reaching onsets. Both V1/V2 and hV6A seem to be causally involved, even if with different roles, in instructing this coordination.
Invalid motor trials
hV6A stimulation impaired the redirection of covert attention
Invalid Motor trials forced participants to automatically disentangle attention after the go onset to bring it to the opposite hemifield. Reaction times of invalid trials were significantly affected by the interaction TMS by redirection side (F(2,32) = 5.55, p = 0.008, partial η2 = 0.26, Fig. 6, individual participants’ data in Fig. 7). This effect was driven by the slower reaction times when attention was redirected leftward, specifically after hV6A stimulation compared with Sham (p < 0.01) and V1/V2 stimulation (p < 0.01), which in turn were not different from one another (p = 0.32). During rightward redirection of attention, reaction times were not affected by TMS (all p > 0.52). As shown in Figure 6, rightward redirections of covert attention during Sham stimulation caused slower reaction times than leftward redirections (p = 0.01), an effect evident also during V1/V2 stimulation (p = 0.002), but impaired specifically during hV6A stimulation (p = 0.43). Overall, the analysis of reaction times of invalid trials revealed a specific involvement of hV6A in disentangling attention from the contralateral hemifield (right in our case). In invalid trials, no significant correlations between pupil responses and performance were observed (nonsignificant main effect of pupil response: β = 0.74 ± 5.75, t = 0.13, p = 0.89; nonsignificant interaction of pupil response and TMS, β = 0.28 ± 2.56, t = 0.11, p = 0.91).
Discussion
The medial PPC is causally involved in attentional orienting and in disentangling attention before reaching
We here find that hV6A is causally involved in reach initiation only if covert attention is allocated to the reaching target. To our knowledge, this is the first study on the PPC where participants were instructed to reach unattended targets. The independent control of attention and reach plan was possible because, in our task, the cue was informative about the location to orient covert spatial attention and the location of the motor plan, which could be the same or different. This design instructed participants to allocate their resources in multiple target locations simultaneously, an ability that was repeatedly demonstrated in humans (Baldauf et al., 2006; Hanning et al., 2018; Schonard et al., 2022). It required additional resources than the control task, as suggested by the increase in pupil size when looking at the informative cue (Fig. 4, black line) compared with the uninformative one (Fig. 4, yellow line), in all the trial types.
Because the spatial congruence of the attention-reach plan was essential for hV6A, one could argue that the activations previously seen in monkey and human V6A during reach planning (Cavina-Pratesi et al., 2010; Breveglieri et al., 2014; Hadjidimitrakis et al., 2014) are only the result of allocation of attention, given that primates naturally allocate attention on reaching targets unless they are forced to behave differently. If this were the case, hV6A should be involved in the process of maintaining attention on a target in the absence of a reach plan. Other studies demonstrated that this is not the case (Capotosto et al., 2013; Ciavarro et al., 2013). Thus, we suggest that the activations during reach planning may not be attributed solely to the allocation of sustained attention. Rather, we suggest that the activations seen in hV6A during reach planning are the result of an allocation of attention that is functional to reach that location. The same phenomenon was observed before saccades (“presaccadic attention”; Li et al., 2021). We thus propose that hV6A has a causal role in “prereaching attention.” In a single-cell monkey study (Breveglieri et al., 2014), the animals were trained to overtly attend a target and, in different trials, to overtly attend and plan a reach to the same targets. The most common modulations of V6A cells in this study were related to both overt attention and reach planning, and this is in keeping with the current results. Therefore, we suggest that hV6A might send information to frontal areas (Gamberini et al., 2009; Tosoni et al., 2015) about the alignment of spatial coordinates of attention and reach planning, and the frontal cortex may use this information to initiate a reach more promptly, resulting in a gain in the system (Fig. 2A,B, black columns). Accordingly, after hV6A stimulation, this gain is lost (Fig. 2B, gray columns). Present findings are also in agreement with studies (Rolfs and Carrasco, 2012; Li et al., 2016, 2019, 2021; Messinger et al., 2021; Schonard et al., 2022) that go against obligatory coupling of attention and motor plans postulated by the premotor theory of attention (Rizzolatti et al., 1987). As the effects of TMS were observed on reaction times, one could argue that the role of hV6A is solely related to the perceptual detection of the go signal. We think we can discard this interpretation, because we have observed a TMS effect only in congruent trials, whereas the go signal detection was required in all the trial types. We also found that hV6A has a causal involvement in redirecting attention in invalid trials (Fig. 6), in agreement with other results where the medial PPC was involved in shifts of attention in “exogenous” Posner paradigms (Vandenberghe et al., 2001; Molenberghs et al., 2007; Ciavarro et al., 2013), and we extend this concept to the “endogenous” attentional orienting mode used here. Differently from other TMS studies (Capotosto et al., 2013; Ciavarro et al., 2013), we show here that the effect of hV6A stimulation is direction-sensitive. Specifically, during Sham stimulation, participants are slower in reach initiation after a rightward redirection of attention than after a leftward redirection (Fig. 6). This “leftward gain” is in line with the right hemisphere's dominance in directing spatial attention (Heilman and Van Den Abell, 1980; Reuter-Lorenz et al., 1990; Benwell et al., 2014) and with lesser efforts required when attending to the left visual field compared with the right (Meyyappan et al., 2023). It is also in line with the concept of “pseudoneglect,” a phenomenon where healthy participants tend to place a bisection marker to the left of the real midpoint on a horizontal line, reflecting a natural trend to attend leftward (Bowers and Heilman, 1980). By stimulating the left hV6A, we found that this “leftward gain” is lost, as suggested by the increase in the reaction time after hV6A stimulation than Sham and V1/V2 stimulation. We thus suggest that hV6A has also a causal role in disentangling the focus of attention from contralateral targets and in the pseudoneglect.
The occipital cortex is causally involved in attention orienting but not in its reorienting
Here, we wanted to investigate the role of hV6A, and the stimulation of bilateral V1/V2 was an active control condition, so the investigation of occipital areas was out of the scope of this research. Nevertheless, our data suggest that V1/V2 are involved, together with hV6A, in reach initiation specifically when movements are directed toward the attended location (Fig. 2B). Like for hV6A, we can rule out that the effects we found are merely perceptual (detection of the go signal), so due to visual masking, a typical effect of V1/V2 stimulation (Amassian et al., 1989). In fact, the effect was restricted to congruent trials. Actually, the concept of V1/V2 as a mere perceptual region has been repeatedly dropped out, because early visual cortices are more active when a stimulus is attended (Brefczynski and DeYoe, 1999; Gandhi et al., 1999; Martínez et al., 1999; Somers et al., 1999), and directing attention even in absence of a visual stimulation activates these regions (Kastner et al., 1999; Ress et al., 2000; Silver et al., 2007; Murray, 2008). It has also been demonstrated that during attentional allocation before saccades, feedback signals are sent from frontal eye fields (FEF) and superior colliculus to early visual cortices to enhance visual processing (Ekstrom et al., 2008; Bisley and Mirpour, 2019; Li et al., 2021; Hanning et al., 2023). Interestingly, the early visual cortex is also involved in reaching, in that reaching directions can be decoded from occipital fMRI activity in sighted (Monaco et al., 2017, 2020) and even in blind (Bola et al., 2023) humans, suggesting that action representation in the occipital lobe is independent of vision and is reach-related. Thus, the role of early visual cortices in orienting attention before reaching shown here is in keeping with these studies. Our data show that, differently from hV6A, V1/V2 is not involved in shifts of attention. Although fMRI studies show that early visual areas are involved in shifts of attention (Dugué et al., 2020; Parisi et al., 2020), it has been reported that they are not causally involved in endogenous attention (Fernández et al., 2023). Thus, our results suggest a strict collaboration between occipital and posterior parietal regions and hV6A seems to deal with higher computational loads. Further studies are required to clarify this interpretation.
hV6A has a specific role in coordinating arousal and reaching initiation
Here, we found that hV6A is not involved in modulating pupil size per se (Fig. 4). This is not surprising, given the absence of direct connections between the medial PPC and the subcortical structures (locus ceruleus included) that control the Edinger–Westphal nucleus (Gamberini et al., 2016, 2020, 2021a,b; van der Wel and van Steenbergen, 2018).
We here demonstrate the existence of a coordinative machinery between arousal and reach initiation, in that larger pupil constrictions predict faster reach initiation (Fig. 5A). This correlation was significant only in trials where spatial attention was allocated to the reaching target and was impaired after hV6A stimulation. The role of hV6A in the coordination of arousal-reaching parallels the same role in pre-reaching attention shown here for congruent trials. Stimulation of the active control site (occipital cortex) led to an opposite coordination, instead of an absence of a coordination, and this demonstrates a functional specialization of hV6A and V1/V2.
A similar orchestration was also found for saccades (Jainta et al., 2011; Wang et al., 2015, 2016, 2017). According to a recent study (Hsu et al., 2021), TMS over FEF impairs the orchestration of pupil size-saccade initiation, revealing the causal role of FEF in this coordination. We here found the functional counterpart of this process in the reaching domain. Again, and in line with the reaction time modulations, a specific role of hV6A was seen only when covert attention was directed at the reaching target.
After V1/V2 stimulation, an opposite correlation between pupil constriction and reaction time compared with Sham stimulation was observed (Fig. 5B). A possible explanation for this effect may be found considering that a negative correlation between pupil size and activation of occipital areas was demonstrated (Bombeke et al., 2016; Lubinus et al., 2022). So, the larger the pupil size, the lower the activation of the visual cortex. We thus can suggest that the interference given by the stimulation to the occipital lobe performed here could lead to a reduction in visual cortex activity, and this may have increased the pupil size, as seen in Figure 5B, where pupil constriction to light has been reduced after V1/V2 stimulation.
Future applications
The neural bases of the interplay between attention and reach planning shown here can inform the development of rehabilitation strategies to address deficits like attention-deficit hyperactivity disorder that often involve impairments in both attention and motor control, ultimately improving functional outcomes.
Footnotes
We thank Dr. Christoph Strauch for his insightful comments on analyses and interpretation of pupil size data. This study was supported by the European Union’s Horizon 2020 research and innovation program under the Future and Emerging Technologies grant agreement No. 951910- MAIA and by Ministero dell’Università e della Ricerca (MUR), PRIN2022-2022BK2NPS.
The authors declare no competing financial interests.
- Correspondence should be addressed to Rossella Breveglieri at rossella.breveglieri{at}unibo.it.
