Tennis expert-novice difference in motion-in-depth perception is associated with early inhibition of invalid attention

Abstract

Perception of motion-in-depth is essential to guide and modify the hitting action in interceptive-dominated sports (e.g., tennis). The purpose of this research was to investigate the effect of tennis expertise on motion-in-depth perception at different cognition processes by two event-related potential (ERP) studies. Specifically, the study explored differences in behavior and electroencephalogram (EEG) characteristics between tennis experts and novices under two different task conditions. Seventeen skilled tennis players (average professional experience: 10.76 ± 2.56 years) and 17 age-matched novices participated in two experiments. Cortical activity (P1, N1, P2, and the later component) and task performance (reaction time, accuracy rate, and error value) were recorded when participants completed a direction-identifying task (Experiment 1) and a prediction-motion task (Experiment 2) while perceiving motion-in-depth. Results demonstrated that P1 latency in the expert group was significantly shorter than the novice group in experiment 1; and in experiment 2, the expert group showed shorter P1 latency than the novice group under the TTC1 (actual time-to-collision is 400 ms) condition. The results indicate that the effect of tennis expertise on motion-in-depth perception was primarily manifested as early visual attentional investment in the features of objects. Experts have the rapid inbibition of invalid attention.

Introduction

Dynamic changes in depth cues, as objects move closer or farther from an observer, lead to the perception of motion-in-depth. Motion-in-depth perception is essential to guide and modify the hitting action in interception-dominated sports (e.g., tennis). During each serve, tennis players need to perceive the movement of the ball as it approaches and as it moves away. Studies have indicated that experts exhibit superior motion-in-depth perception than novices^1,2. This capability correlates with improvement in one-handed catching³ and visual-visuomotor ability⁴. According to the study of Gao et al.⁴, athletes in ball games performed significantly better than non-athletes in the tasks of manual interception and predicting the depth of moving objects, and had better depth dynamic visual acuity of dominant eyes than non-athletes. Despite evidence that expertise in interception sports enhances motion-in-depth perception, the specific cognitive processes involved remain unclear. Expert athletes use the visual system in a different, and potentially more effective, manner when scanning and extracting information on display⁵. These cognitive characteristics can be reflected in brain activity.

Existing research has described the temporal and spatial characteristics of brain activity when perceiving motion-in-depth, using planar 2D paradigms including optical flow⁶, random-dot patterns⁷, and checkerboard patterns^8–10. However, there is a considerable difference between these paradigms and actual visual experience. Wang and Yao¹¹ employed a 3D paradigm that is closer to human visual experience and combined it with ERP techniques to investigate the effect of the object’s size, direction, and speed on the perception of motion-in-depth. They found that N220, P140, and P300 might be ERP components closely related to motion-in-depth perception¹¹.

The ability of motion-in-depth perception involves not only the rapid extraction of spatial information such as direction, trajectory and speed, but also the integration of spatial and temporal characteristics, such as the accurate judgment the point of ball landing. According to the Information Processing Theory¹², the human brain undergoes a process of encoding and decoding information, transitioning from complexity to simplicity. Tennis players first need to perceive the direction and speed of the object about to collide, then estimate the relative distance to the object, and use advanced cognitive activities to calculate temporal information before hitting the ball. This cognitive process involves an indirect transformation from environmental stimuli to perception, with the former primarily concerning the judgment of motion direction perception, and the latter focusing on the judgment of collision time with approaching objects. Therefore, it is of great value to explore the specific cognitive processes underlying the expertise effects in motion-in-depth perception across multiple levels. ERP components related to motion-in-depth perception direction judgment include P1, N1 and P2^13,14. ERP components related to the judgment of time-to-collision (TTC) of approaching objects include P1, N1-frontal, N1-occipital and the late slow component¹³. Among them, P1 is related to some primary features of visual stimuli and inhibition of unconscious sensory bias¹⁴. The occipitotemporal N1 reflects a certain discrimination process, and this component is not affected by reaction correlation or preparation process¹⁵. In the experiment that requires appropriate responses based on stimulus appearance, P2 is considered to be a closely related ERP component to working memory¹⁶. Additionally, the late slow component reflects the brain’s encoding and processing of visual-spatial information and are considered to play a key role in the consolidation and retention of memory¹⁷.

Extensive studies have shown that cognitive abilities can be improved with individualized adaptive training, and these improvements are reflected in more efficient neural networks^18–20. In complex sports scenarios, high-level athletes have a lower perceived anticipation time compared to low-level athletes and non-athletes, with this advantage manifesting at various stages of the sport. Fencing experts evoked lower P1, N1, and N2 amplitudes in specific brain regions when making predictions, and shorter latencies for N2 and P3 in specific brain regions²¹. The technical anticipation process of experts is characterized by efficient visual search, fast recognition, and less mental energy used. Research on tennis players’ anticipation of serve direction shows that 42 ms before the racket contacts the ball, high-level tennis players can predict the serve direction more accurately than beginners; however, 336 ms after the ball contacts the racket, there is no significant difference in accuracy between the players and beginners²². Based on previous research, the expertise effect of motion-in-depth perception is reflected in the early invocation and allocation of cognitive resources. Tennis experts and novices invested different attention resources in different directions of perception²³, attention resources are invested earlier in experts rather than novices²⁴. Gray integrates relevant research from the fields of psychology and sports science, proposing the view that “an athlete’s perception depends on the interaction between the physical characteristics of the objects in the sports environment and the athlete’s own abilities.” Gao, et al.⁴ indicated that visual ability (including depth motion perception ability) can predict skills related to visual perception. Based on the above research, it is inferred that specialized cognitive resources are a reason for the cognitive expertise advantage of experts. However, due to the influencing factors such as the rotation of stimuli²³, movement speed²⁴, and emotions²⁵ in the above-mentioned studies, we cannot directly compare different processes of motion-in-depth perception to investigate the effect of expertise.

To investigate the effect of tennis expertise on motion-in-depth perception at different cognition processes, two experiments with matching visual stimulus parameters (shape, starting position, and speed) were conducted. Experiment 1 explored the effect of tennis expertise on motion-in-depth direction discrimination, and Experiment 2 mainly explored the effect of tennis expertise on TTC judgment. Both experiments involved primary information extraction during motion-in-depth perception. Whereas Experiment 1 focused on discrimination and recognition, and Experiment 2 focused on the estimation of TTC from motion information. Based on previous studies, we hypothesized that compared to the novices, tennis experts would have higher accuracy in judging direction but have no difference in reaction time (Experiment 1) (H1a). We also hypothesized that compared to the novices, tennis experts would have faster cognitive resource extraction and less cognitive resource engagement. This characteristic is particularly reflected in the rapid processing motion-in-depth information, and in the TTC estimation process, it is more reflected in attention resource saving. Specifically, compared to novices, experts would show shorter latency as well as smaller amplitude at P1, N1, and P2 in a direction-identifying task (Experiment 1) (H1b); shorter latency as well as smaller amplitude at P1 and N1, and smaller amplitude at the later component in a prediction-motion task (Experiment 2) (H2).

Experiment 1

Participants

A total of 17 varsity tennis players (7 males, 10 females, aged 20.8 ± 2.25 years) were recruited as expert participants. Their average years of professional experience was 10.76 ± 2.56 years. A total of 17 college students (8 males, 9 females, aged 21.2 ± 2.14 years) were recruited as novice participants. The expert participants satisfied all of the following inclusion criteria: (1) qualified as a National Player in second grade or above; (2) had five or more years of professional training experience; (3) practiced more than three times a week for two or more hours in the last 2 years. The novice participants took an 18-week tennis course (1.5 h per week) but did not meet any of the criteria for being a tennis expert. All participants had normal or corrected-to-normal vision and were right-handed. They were informed about the testing protocol and signed a written consent form. The study was approved by the Institutional Research Ethics Committee of the institution with which the first author was affiliated.

Design

The study employed a 2 (Group: expert vs. novice) × 2 (Direction: approaching vs. receding) mixed experimental design. Group was the between-subjects factor and directions were the within-subjects factor. The dependent variables were reaction time, accuracy rate, peak amplitude, and latency of P1, N2, and P2.

Tasks, materials, and measures

Following previous research, we adapted the experimental task of non-collision motion-in-depth perception based on the existing studies^9,26. Three videos were generated by Autodesk 3ds Max 2010. Two videos were corresponding to the two experimental conditions of “approaching” and “receding”. The third video was non-target stimulation (i.e., “stationary”) and required no reaction.

In the videos of experimental conditions, a sphere with an initial diameter of 4.91° appeared in the center of the screen and continually to expanded or contracted for 1000ms until the sphere reached a diameter of 8.17° or 1.64°. In the stationary video (i.e., the non-target stimulation video), the sphere was kept stationary at a diameter of 4.91° until the 1000 ms video ended.

Speed-accuracy task

Participants were asked to identify the motion direction of the sphere as quickly as they could and to press the corresponding key according to the direction.

EEG recording and offline process

EEGs were recorded from Ag/AgCI electrodes mounted in a 64-channel Brain Cap (Brain Products, Germany) based on the international 10–20 system by two 32-channel amplifiers with a sampling frequency of 1000 Hz (Brain Products, Germany). All sites were referenced to FCz and offline re-referenced to the average of 63 electrodes (all except the EOG electrode). The electrode impedance for EOG and EEG electrodes was less than 5 kΩ. Ocular artifacts were removed from EEGs using the Ocular Correction Independent Component Analysis implemented in Brain Products software (Brain Vision Analyzer 2.1). EEGs were filtered with a 0.01–30 Hz bandpass and segmented beginning 200 ms prior to the onset of targets (i.e., the approaching sphere) and lasting for 1000 ms. All epochs were baseline-corrected with respect to the mean voltage over the 200 ms preceding the onset of targets, and epochs with amplitude beyond ± 75 µV were rejected as artifacts. The average ERPs were computed based on no less than 45 trials per condition.

Procedure

All testing was conducted in a dimly lit room to reduce distraction and allow for valid testing. Each participant was seated 70 cm from a 19” LCD screen. All stimuli were presented on a black ground. Before the data collection, participants were given 12 practice sessions with feedback about their reaction time and accuracy.

Each trial was presented in the following order: Fixation point (randomly 200–400 ms) → buffer (400 ms) → randomly one of the three videos → inter stimulus interval (Fig. 1). Participants completed the speed-accuracy task by quickly indicating the direction as quickly as they could (if the participant did not respond in 1000ms, the video ended and it would be considered as a wrong answer), and the sphere was moving via the corresponding key according to the movement direction of the ball (“approaching” for the “J” key, “receding” for the “F” key, “stationary” for no reaction). The entire task included 180 trials, with 60 approaching trials, 60 receding trials, and 60 stationary trials. All trials were evenly divided into two blocks with random presentation. The whole experiment took about 9 min, including a 1-min break between the two blocks.

Fig. 1.

Diagram of one trial in Experiment 1.

Data analysis

The mean amplitudes (µV) and mean latencies (ms) of P1, N1, and P2 were analyzed in accordance with grand-mean ERP topographies and the literature^10,23,27. The electrode sites and time windows where amplitudes and mean latencies were calculated were as follows: P1 was measured at PO7 and PO8 (90–130 ms); N1 at PO7 and PO8 (150–200ms); and P2 was measured at Oz (200–250 ms).

All statistical analyses were performed with IBM SPSS Statistics (version 19), with the level of significance set at p < 0.05. We performed a series of 2 × 2 repeated measures ANOVAs. Partial eta-squared (η²p) was used as the measure of effect size.

Results

The accuracy rate and reaction time between the two groups are shown in Table 1. The average accuracy rate of the two groups of participants under all conditions exceeded 95%, and the standard deviation was within 5% points.

Table 1.

Behavioral measures in the judging direction paradigm of two groups.

Group	Tennis Expert		Novice
	M	SD	M	SD	F	p
ACC of approaching	0.96	0.04	0.96	0.03
ACC of receding	0.95	0.04	0.97	0.04
Direction					0.00	1.00	0.00
Group					0.14	0.71	0.00
Direction × group					1.80	0.19	0.05
RT of approaching	441.20	57.52	453.41	26.74
RT of receding	440.50	67.78	464.75	32.28
Direction					4.42	0.04*	0.12
Group					0.54	0.47	0.02
Direction × group					2.10	0.16	0.06

ACC Accuracy rate, RT  Reaction time (ms).

*p < 0.05.

Repeated measurement ANOVA showed that there were no significant differences in the accuracy of approaching and receding direction judgments between tennis experts and novices (ps > 0.05). The average reaction time of the expert group was not significantly different from that of novice.

According to the characteristics of the total average waveform figure (Fig. 2A), the peak detection range of each component was confirmed: P1 (90–130 ms), N1 (150–200 ms), and P2 (220–280 ms). As presented in Fig. 2B, P1 and N1 mainly appeared in the temporal occipital areas on both sides. P2 appeared in the occipital area and extended to both sides.

Fig. 2.

Different neural activation between tennis expert and novice. (A) Total average waveforms at the electrode points PO7, PO8, and Oz. (B) The instantaneous topographical map of P1, N1, and P2 in the ERP total average map with 106 ms, 163 ms, and 232 ms representing these three components respectively.

Repeated measures ANOVAs were performed respectively for the amplitude (µV) and latency (ms) at P1 and N1 at PO7 and PO8 electrode points; P2 at PO7, PO8, and Oz electrode points. Group (expert vs. novice) was the between-subjects variable, and movement direction (approaching vs. receding) was the within-subjects variable. Results that achieved a statistically significant effect (p < 0.05) are shown below (values represent mean ± SE).

P1: The analyses of P1 latency revealed a significant main effect of group on PO7, F_{(1, 32)} = 5.20, p = 0.029, = 0.12. The P1 latency of the expert group (106.35 ± 2.58 ms) was significantly shorter than that of the novice group (114.68 ± 2.58 ms).

Experiment 2

Participants

The same expert and novice participants took part in Experiment 2 at least one week after they finished Experiment 1. The two experiments were completed within one month.

Design

The study employed a 2 (Group: expert vs. novice) × 2 (TTC: 400 ms vs. 800 ms) mixed experimental design. Group was the between-subjects factor and TTC was the within-subjects factor. The dependent variables were reaction error value, the peak amplitude and latency of P1, N1, and the average amplitude of the later component.

Tasks, materials, and measures

Four videos were generated by Autodesk 3ds Max 2010. In each video, a 3D checkerboard sphere (initial diameter: 1.64°; luminance for white elements: 120 cd/m²; mean luminance: 55 cd/m²; contrast: 100%) appeared in the center of the screen and continually expanded at a constant speed. The collision reference was composed of two gray lines perpendicular to the horizontal axis of the screen and with equal distance to the vertical axis.

Referring to previous studies on collision time estimation^28,29 and the characteristics of individual visual processing³⁰, this study establishes two collision time estimation conditions at 400ms and 800ms. Two videos set in the practice block were used to demonstrate the collision under the two experimental conditions (TTC1: 400 ms vs. TTC2: 800 ms) and familiarize the participant with the operation of the task. In the demonstration video, the expansion of the sphere lasted 1400 ms (or 1800 ms) until the boundary of the sphere touched the collision reference (i.e., the two gray lines) with a terminal diameter of 8.50° (or 10.44°). The distance between the two gray lines was exactly equal to the diameter of the sphere at the time of the collision.

The other two videos were used during the formal experiment. In the TTC1-experimental video, the sphere expansion lasted 1000 ms until its diameter reached 6.54° and it disappeared. The distance between the two gray lines of the collision reference was 8.50° (representing a remaining time to the collision of 400 ms). In the TTC2-experimental video, the sphere’s expansion lasted 1000 ms until its diameter reached 6.54° and it disappeared. The distance between the two gray lines of the collision reference was 10.44° (representing a remaining time to the collision of 800 ms).

Prediction-motion task

Participants were asked to view the videos of the approaching sphere and to press the “J” key at the moment that coincided temporally with their prediction of when the sphere would arrive at the collision reference point.

EEG recording and offline process

Consistent with experiment 1.

Procedure

Before data collection, participants familiarized themselves with the key operation and understood the operational definition of collision by practicing a 20-trial block called the collision demonstration section. The operational definition of collision in this study was the moment when the sphere contacted the reference plane composed of two gray lines. After the collision demonstration section, participants familiarized themselves with the task in a 12-trial block called the practice section (every 6 trials for TTC1 and TTC2). The whole process of stimulus presentation and operation in this section would be the same as the formal experiment, except that feedback would be provided after the response. Figure 3 shows the flow chart of stimulation presentation in the collision practice section.

Fig. 3.

Flow chart of stimulation presentation in collision practice demonstration. (Note. In the 1000 ms feedback frame, the displayed number “123” would indicate that the participant’s reaction moment was 123 ms later than the actual collision moment.)

After the practice section, the formal experiment began and only data from the formal experiment would be included in the analysis. Each trial was presented in the following order (Fig. 4): fixation point (randomly 200–400 ms) → collision reference (400 ms) → TTC1 or TTC2 experimental video (1000 ms) → reaction interval (unlimited time, the end after the participant pressed the button) → interstimulus interval (randomly 200–400 ms). Participants completed the prediction motion task in which a 3D sphere expanded in size over 1000 ms (simulating visual approaching) and then disappeared. After the sphere disappeared, the participants had to imagine the stimulus continuing to approach and judge when it would have contacted with the collision reference. After they pressed the “J” key to make their judgment, there would be an interval between stimuli (lasting randomly 800–1200 ms) before the next trial began. In the formal experiment, the entire task included 120 trials and the two conditions were 60 trials each. All trials were divided into two blocks with a random presentation. The whole experiment took about 9 min, including a 1-minute break between the two blocks.

Fig. 4.

Diagrammatic sketch of one trial in Experiment 2.

Data analysis

All statistical analyses were performed with IBM SPSS Statistics (version 19), with the level of significance set at p < 0.05. We performed a series of 2 × 2 repeated measures ANOVAs (RM-ANOVAs), each with one between-subjects factor (Group: expert vs. novice) and one within-subjects factor (TTC: 400 ms vs. 800 ms). We express data as mean ± standard error (SE). Partial eta-squared (η²p) was used to determine effect size.

Behavioral data were filtered, merged, and preprocessed through E-Data Aid. After eliminating outliers (defined as data points falling three standard deviations above or below the mean), we obtained both the TTC error value (i.e., the time from the moment the sphere disappeared to the moment that the participant pressed the “J” key minus the actual time-to-collision) of each participant under each experimental condition.

The mean amplitudes (µV) and mean latencies (ms) of P1 and N1 were analyzed in accordance with grand-mean ERP topographies and relevant literature¹³. The electrode sites and time window for calculated mean amplitudes and mean latencies were calculated at the following locations: P1 was calculated at PO7 and PO8 (100–150 ms); N1 at PO7 and PO8 (160–220 ms); and the later component at Fz and FCz (400–1000 ms).

Results

The reaction time and error value of the collision time between the two groups are shown in Table 2. The reaction time of the expert group and the novice group both exceeded 1400 ms (TTC1) and 1800 ms (TTC2), respectively. Results showed that the main effect of time-to-collision was significant (F_{(1, 32)} = 294.61, p < 0.001, = 0.90). The mean reaction time of the TTC1 trials (1640.49 ± 26.45 ms) was significantly shorter than the mean reaction time of the TTC2 trials (1962.54 ± 39.69 ms). This confirms that the experimental stimulus design achieved the expected differentiation effect.

Table 2.

Behavioral measures in the prediction-motion paradigm of two groups.

Group	Tennis expert		Novice
	M	SD	M	SD	F	p
ERR of TTC1	322.55	143.06	269.82	157.20
ERR of TTC2	323.10	189.58	275.78	157.78
TTC					0.04	0.84	0.00
Group					0.88	0.36	0.03
TTC × group					0.27	0.87	0.00
RT of TTC1	1705.26	151.61	1633.31	177.52
RT of TTC2	2059.57	230.99	1988.58	215.10
TTC					294.61	0.00**	0.90
Group					1.24	0.27	0.04
TTC × group					0.00	0.98	0.00

ERR  Error value, TTC Time-to-Collision, RT Reaction time.

**P < 0.01.

Next step, we conducted separate 2 × 2 repeated measures ANOVAs using error value as the dependent variables. Results showed that there was no significant main effect of either group or TTC (ps > 0.05). Examination of the standard deviation error value revealed that even within the same group, the differences between individuals were relatively significant.

Figure 5A shows the ERP total average waveform figure at the electrode points PO7, PO8, Fz, and FCz. According to the characteristics of the total average waveform figure, the peak detection range of each component was confirmed: P1 (100–150 ms), N1 (160–220 ms), and the later component (400–1000 ms). As presented in Fig. 5B, P1 mainly appeared in the occipital area and extended to both sides. N1 appeared in the temporal occipital areas on both sides. There was also a significant late slow component in the frontal and central areas. Repeated measures ANOVAs were performed for the peak amplitude (µV) and latent period (ms) respectively at P1, N1 at PO7 and PO8 electrode points, and the later component at Fz and FCz electrode points.

Fig. 5.

Different neural activation between tennis expert and novice. (A) Total average waveforms at the electrode points PO7, PO8, Fz and FCz. (B) The instantaneous topographical map of P1, N1, and the later component in the ERP total average map with 129 ms, 195 ms, and 609 ms representing these three components respectively.

Results that achieved a statistically significant effect (p < 0.05) are shown below (values represent mean ± SE).

P1: The analyses of P1 amplitude revealed that the main effect of the TTC was significant, F_{(1, 32)} = 11.49, p = 0.002, = 0.15. The P1 amplitude for TTC1 (3.08 ± 0.47 µV) was significantly larger than that of TTC2 (2.31 ± 0.54 µV). Additionally, repeated-measure ANOVA carried out on amplitudes (F_{(1, 31)} = 4.33, p = 0.046, = 0.12) and latencies (F_{(1, 31)} = 5.51, p = 0.025, = 0.15) showed that there was a significant interaction between TTC and group. Simple effect analysis found that the P1 latency of the expert group (118.82 ± 3.24 ms) was significantly shorter than that of the novice group (130.76 ± 3.24 ms) at TTC1. And The amplitude of the expert group on TTC1 (0.04 ± 0.50 µV) was significantly larger than TTC2 (– 0.57 ± 0.58 µV).

Discussion

The current study investigated the effect of tennis expertise on motion-in-depth perception across different cognitive processes using two ERP experiments. In Experiment 1, participants quickly identify the stimulus features in the perceptual process and make an accurate response after encoding and comparing the information. Motion-in-depth perception mainly involves the cognitive process of quickly allocating attention resources for information extraction and recognition. The hypothesis H1a was not fully supported. Consistent with prior studies, there were no performance differences between experts and novices in a pure direction judgment task^23,24. However, the results of the accuracy rate argue against H1a, there were no significant differences in the accuracy of direction judgments between tennis experts and novices. The average accuracy rate for both groups under all conditions exceeded 95%, with standard deviations within 5% points. This suggests a ceiling effect on the accuracy, and this might weaken the differentiating effect between groups.

According to the behavioral results of Experiment 2, participants overestimated the collision time compared to the actual collision time, with an average error between 269 and 323 ms. This is different from the results of previous studies using the prediction-motion paradigm^31,32. In post-experiment interviews, participants reported that before the stimulus disappeared, they tried to perceive the sphere’s motion speed and reacted accordingly. However, later on, some based their responses on a certain rhythm, and a few even counted fixed intervals after the sphere disappeared These strategies might lead to significant individual differences in how the participants approached the task.

Based on the ERP results of Experiment 1, there was no significant difference in the amplitude of P1, N1, and P2 between the expert group and the novice group, and no support for the hypothesis H1b. Only the result of P1 latency in the left occipitotemporal region supported H1b. Like Experiment 1, there was no significant difference in the amplitudes of P1 and N1 between the expert group and the novice group, not supporting H2. Only the P1 latency result in the left occipitotemporal region under the TTC1 condition partially supported H2.

In trial-by-trial cueing tasks, the evoked P1–N1 components reflect attentional mechanisms of early sensory gain control or amplification³³. P1 and N1 are generally associated with early processing of physical stimuli. There was no significant difference in P1 amplitude between the expert and novice groups, against H1b. This indicated that both experts and novices invested the same attention resources in perceiving simple visual stimuli. In Experiment 1, the sphere had the same initial size in both approaching and receding conditions, which differs from previous studies^23,24. The same initial size was used to prevent participants from using size rather than motion to judge direction, ensuring similar attentional resource investment across participants. P1 is related to some primary features of visual stimuli and inhibition of unconscious sensory bias¹⁴. Supporting H1b, the P1 latency in the left occipitotemporal region was shorter in the expert group than in the novice group, indicating that experts allocated attentional resources for primary feature recognition faster than novices.

In Experiment 2, the P1 amplitude for TTC1 was significantly greater than that for TTC2 in the expert group, with a notable difference in P1 amplitude under TTC1 conditions compared to TTC2. It indicated that individuals invested more attention resources in the early stages of events with relatively short TTC, which is more urgent. Under the TTC1 condition, the P1 latency also exhibited a similar expertise effect as in Experiment 1.

The average reaction time of Experiment 1 was around 450 ms. In Experiment 2 the TTC1 condition, it was setting 400 ms from the ball disappeared to its contact with the collision reference. Therefore, participants also need to quickly and accurately determine TTC. It is speculated that in tasks requiring rapid responses, the expertise effect reflects that experts can attend to the primary features of visual stimuli earlier than novices.

In this study, the negative component after P1 was mainly distributed in the bilateral temporo-occipital region, and the latency range was between 150 and 210 ms. Based on a series of studies, Hillyard³³ concluded that this visual component is related to the direction of attention: decreased P1 amplitude in the bilateral occipital regions reflects inbibition of non-attended positions, while increased N1 amplitude reflects enhancement of attended positions and is related to target detection facilitation. In Experiment 1, P1 latency was shorter in the expert group than in the novice group, but the subsequent N1 latency did not maintain this expertise effect. Therefore, it can be inferred that experts’ cognitive advantage in the process of motion-in-depth perception in general situations is reflected in the faster inhibition of ineffective attention direction, rather than the enhancement of target detection attention direction. In Experiment 2, all parameters of the sphere under the two TTC conditions were the same, and there was no difference in target position detection before the spheres disappeared. Therefore, there was no difference between the two TTC conditions in occipitotemporal N1. In previous ERP studies related to prediction or identification, N1 amplitude in the expert group was significantly larger than in the novice or non-athlete group^34,35. Therefore, it is suggested that in experiments with simpler stimulus scenarios, experts would have no significant discrimination or recognition advantage in collision time estimation.

The occipital P2 is associated with the recognition of the target stimulus and reflects the integrated processing of input information. Against H1b, there were no significant group differences in the amplitude and latency of P2 in this study. It is speculated that group differences in pattern recognition decrease when the task is simplified, and it could not reflect the experts’ speed advantage in information recognition during complex motion situations²³.

According to the ERP data results, there was no significant difference in the average amplitude of the late slow component between the expert group and the novice group, not supporting the expertise effect of attention resource investment savings in H2. In Experiment 2, participants needed to make subsequent TTC judgments based on the velocity information of constantly approaching objects. The late negative components that appear between 400 ms and 1000 ms after P1 and N1 reflected this cognitive process. Multiple studies have found that memory encoding is associated with a negative slow wave (SW), and that the scalp distribution of this wave varies by sensory channel and type of information^36–38. Barcelo³⁹ pointed out that this long persistent slow wave is an ERP component specifically responsible for encoding visual spatial information and reflecting working memory. In this study, the SW related to memory coding was not affected by tennis expertise, and there was no significant difference between different TTC conditions. This may indicate that individuals with different levels of tennis expertise invoke the same cognitive resources in simple task scenarios with a single stimulus pattern.

A limitation of our study is the fact that the task of motion-in-depth excluded other available cues in the experimental context, and it’s possible to have differences from the real sports context. Although it could examine the pure process of motion-in-depth perception, the differences between expert and novice might be narrow. Additionally, we recruited participants who were college varsity athletes, and the expertise effect might be not significant enough. In the future, further experiments could find participants with higher levels of athletic proficiency to compare with novices. Moreover, more variables involved in sports scenarios can be added to assess whether the expertise effect is more pronounced in the perceptual phase of discrimination, such as time estimation.

Conclusion

This research combined perception before direction judgment and collision time estimation aims to delineate the role of tennis expertise on motion-in-depth perception through a multifaceted exploration of cognitive processes. In summary, the findings of this research demonstrate that the effect of tennis expertise on motion-in-depth perception is related to attention to information extraction and recognition. The cognitive advantage of experts lies in their rapid inhibition of invalid attention, which is reflected in the early visual attentional investment in the features of objects. These findings enrich the reservoir of psychological training in interceptive-dominated sports.

Author contributions

X.W. and C.W. contributed equally to this manuscript and should be considered co-first authors. X.W. conducted the research and investigation process, in particular conducting the experiments. C.W. conducted the data research and was a major contributor to the writing of the manuscript. Y.S. managed and co-ordination responsibility for the planning and execution of the research activities. C.Q. supervised and led the planning and execution of the responsible research activities. All authors read and approved the final manuscript. All authors read and approved the final manuscript.

Funding

The present study was supported by a grant from the National Natural Science Foundation of China to the corresponding author of the article [Grant number 31671161] and the Doctors’ Started Funded Project of Hanshan Normal University to the first author [Grant number: QD20190912].

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Ethical approval and consent to participate

This study was approved by the Ethics Committee of Wuhan Sports University (2023019), which was conducted in accordance with the Declaration of Helsinki. Consent to participate was obtained from each participant.