Successful social interaction requires not only the accurate perception of actions performed by other actors, but also the accurate prediction of what actions they are capable of doing. This skill is required both for coordinating joint actions, such as performing a basket toss in cheerleading, and for predicting other actors’ capabilities for performing individual actions, such as blocking a shot in basketball. Thus, a crucial component of perceptual development is learning to both distinguish the action boundaries of others from our own and distinguish the action boundaries of one actor from another.

The evolution of the human visual system has resulted in perceptual sensitivity to human motion (Johansson, 1973). This sensitivity may facilitate action prediction. Visual access to movement kinematics allows more accurate prediction of another person’s action boundaries, and experience in playing a sport improves the accuracy of these predictions. Here, we evaluate the role of different types of movement in perception of another person’s action boundaries, as well as the impact of experience playing basketball or soccer.

Perceiving affordances for others

Gibson (1979/1986) proposed that we perceive the world around us by detecting information that specifies possibilities for performing an action. These real opportunities for action— which he termed affordances— are specified by informational variables: structured energy patterns to which animals are sensitive. Humans successfully perform actions because they perceive affordances that change from moment to moment on the basis of environmental context and the action capabilities of the observer. Humans can perceive affordances for themselves, as well as for others, and are remarkably accurate at doing so (Carello, Grosofsky, Reichel, Solomon, & Turvey, 1989; Fajen, Riley, & Turvey, 2009; Gibson, 1979/1986; Mark, 1987; Warren, 1984). Consider, for example, the affordance of reaching-while-jumping (RWJ); the perception of the maximum height one can reach overhead by performing a vertical jump. Humans can perceive this affordance for another actor without ever seeing the actor jump (Ramenzoni, Riley, Davis, Shockley, & Armstrong, 2008; Ramenzoni, Riley, Shockley, & Davis, 2008; Weast, Shockley, & Riley, 2011). Presumably, this ability requires that information about affordances exists in structured ambient energy arrays that specify the animal–environment system and that perceivers can detect that information.

Role of movement kinematics

Movement kinematics provide rich information for identifying many characteristics of a moving actor (Abernethy, Gill, Parks, & Packer, 2001; Dicks, Button, & Davids, 2010; Johansson, 1973; Troje, 2002). These can be identified even when limited information is available to the perceiver (e.g., when biological motion of the human body is displayed as point-light videos, ambiguous displays of isolated points of light corresponding to joint centers on the body that disambiguate once movement occurs; Johansson, 1973). While these displays lack the detail available in live or filmed movement, the essential kinematic information available is sufficient for viewers to identify the sex (Kozlowski & Cutting, 1977), age (Montepare & Zebrowitz-McArthur, 1988), and emotional state (Dittrich, Troscianko, Lea, & Morgan, 1996) of the moving actor; to differentiate actors as team-mates or strangers (Steel, Adams, & Canning, 2007); to distinguish one’s own walking patterns from those of others (Jokisch, Daum, & Troje, 2006); and to estimate the perceived velocity of a walking actor (Groner & Schollerer, 2005).

Movement plays an important role in the perception of certain types of affordances for other actors. Perception of affordances involving the force-production capabilities of others requires information about the forces an actor can produce. Because motion is lawfully related to the forces that generated that motion, information about those forces is available for perceivers in movement kinematics (the kinematic specification of dynamics [KSD] principle; Runeson, 1977/1983, 1994). Accurate perception of RWJ, therefore, requires perceptual information about the actor’s capacity to produce vertical propulsive force, which may be available in the kinematics of the actor’s movements during certain other actions. Perception of RWJ for another actor improves after seeing the actor walk or squat but does not improve after seeing the actor twist (Ramenzoni, Davis, Riley, & Shockley, 2010; Ramenzoni, Riley, Davis, et al., 2008; Weast et al., 2011). Movement kinematics during walking and squatting may contain information about the actor’s capacity to produce a different but related action—jumping to reach an object.

Affordance perception in sports

Skilled athletes may develop enhanced perceptual abilities, possibly because they are attuned to different perceptual information than are novices (Abernethy et al., 2001; Hove, Riley, & Shockley, 2006; Oudejans, Michaels, Bakker, & Dolné, 1996). In dynamic environments such as those encountered in team sports, athletes must perceive others’ movement possibilities very quickly, not only for familiar teammates, but also for new opponents (Fajen et al., 2009). A superior ability to perceive the action capabilities of others has been found for athletes in several sports. These include perceiving another’s maximum jumping height and predicting ability to successfully make a free shot, as well as detecting deceptive ball-passing intentions, in basketball (Aglioti, Cesari, Romani, & Urgesi, 2008; Sebanz & Shiffrar, 2009; Weast et al., 2011); predicting the type and length of balls bowled in cricket (Müller, Abernethy & Farrow 2006); estimating the direction and landing position of a serve in tennis and volleyball (Huys et al., 2009; Starkes, Edwards, Dissanayake, & Dunn, 1995); and predicting the direction and magnitude of a stroke in badminton and squash (Abernethy, 1990; Abernethy et al., 2001). With respect to kinematic information, Weast et al. (2011) found that perception of RWJ for another actor improves more so for basketball players than for nonplayer controls after seeing the actor walk.

Present research

The central theoretical questions addressed by this study concern the nature of the information observers use to perceive another agent’s action capabilities and the way observers’ long-term perceptual–motor experiences shape their sensitivity to information about domain-relevant affordances for other actors. Information-based approaches to action understanding (e.g., Ramenzoni, Riley, Shockley, & Davis, 2008) and joint action (Marsh, Richardson, & Baron, 2006) broadly predict, on the basis of the KSD principle (Runeson, 1977/1983, 1994), that task-relevant kinematic information will be sufficient to support accurate perception of actions afforded for another actor. Previous studies that support this prediction (Ramenzoni et al., 2010; Ramenzoni, Riley, Davis, et al., 2008; Weast et al., 2011) have, however, used live actors and, thus, have not adequately controlled for pictorial cues and other nonkinematic information sources that could influence perceptual reports. These studies are unable to rule out the possibility that inferential mechanisms based on cues that relate only contingently to an action, such as maximum jumping distance or height, are the primary constraint on perception of affordances for other actions, as opposed to the underlying dynamic properties of actors that relate lawfully (rather than contingently) to the actions in question. Ecologically inspired theories of perceptual learning (Jacobs & Michaels, 2007; Michaels & Carello, 1981) furthermore predict that long-term experience in an activity such as a sport will result in attunement to the relevant perceptual variables that specify the actor’s action capabilities in a given environmental scenario. Attunement refers to reliance on informational variables that unambiguously specify the to-be-perceived property, on one extreme of a continuum, whereas a lack of attunement could result in the observer relying on informational variables that do not relate in a one-to-one, unambiguous fashion to the to-be-perceived property (Fajen et al., 2009; Gibson, 1979/1986; Jacobs & Michaels, 2007; Michaels & Carello, 1981). Attunement develops over the course of learning and development as an agent has the opportunity to discern the relations between perceptual variables and the agent’s action capabilities.

The first aim of this study was thus to evaluate whether observers can accurately perceive the action boundaries of other actors (for both a vertical and horizontal jump) after viewing point-light representations of the actors’ movements, and whether the type of movement viewed (related/not-related to action) influences perceptual accuracy. Viewing an actor’s live movements improves perception of RWJ if the movement is related to jumping (Ramenzoni et al., 2010; Ramenzoni, Riley, Davis, et al., 2008; Weast et al., 2011). No prior research has evaluated whether this improvement occurs after viewing jumping-related movements of only the joint centers of point-light actors, nor is it known whether this improvement occurs for other affordances. We expected reports for both maximum vertical jumping height (RWJ) and horizontal long-jumping distance (LJ) to be more accurate after viewing point-light videos of actors performing movements related to these affordances (squatting and walking), when compared with movements unrelated to these affordances (twisting and balancing on one leg).

The second aim was to determine whether sports experience improves accuracy in perceiving the action boundaries of other actors both for actions relevant and for those not relevant to the sport. Long-term training in a sport improves perception of affordances relevant to the sport (Abernethy, Neal, & Koning, 1994; Higuchi et al., 2011; Weast et al., 2011), but it is not clear whether this enhanced ability persists for action-scaled affordances that are not relevant to the sport in which the athlete competes. Basketball players have much experience observing others reaching overhead while jumping, since this is a common strategy for retrieving jump balls when beginning/resuming play, performing jump shots, or jumping to reach a passed or rebounded ball, as well as blocking shots/passes to opponents. This action is not as frequently observed in soccer, given that even goalies usually do not have to jump to their maximum height to block a shot. Neither basketball nor soccer players are likely have greater experience than nonplayer controls in observing the standing long-jump, since this action is unlikely to be used in playing either sport.

Method

Participants

Twenty-seven (cf. Weast et al., 2011) male undergraduates at the University of Cincinnati participated for course credit (mean age = 21.58 years ± 4.30). Participants included nine basketball players (all had played on an organized basketball team within 2 years prior to the study), nine soccer players (all had played on an organized soccer team within 2 years prior to the study), and nine nonplayer controls (none had played on a basketball or soccer team).

Materials

Point-light stimulus videos were created by collecting kinematic movement data from three models, each with different maximum RWJ heights and LJ distances (Table 1). Motion data were acquired using an Optotrak Certus motion-capture system (Northern Digital Inc., Waterloo, ON). Fourteen infrared markers were affixed to the ankle, knee, hip, wrist, elbow, and shoulder joints on each side of the body, as well as on either side of the head at the temples (skin-tight Lycra suits were worn to allow proper marker placement). Models performed four actions while their motion was recorded, including a one-leg standing balance (standing unassisted on the right leg only), twisting (standing while twisting side-to-side at the waist), squatting (standing while lowering the torso until the hip and knee joints horizontally aligned, then returning to a normal standing position, keeping the back straight and the arms at the side of the body), and walking (on a treadmill at a comfortable speed). Each action was performed for 60 s, producing a total of 12 point-light videos, 1 of each action type for each model. Videos were displayed on a 127-cm (50-in.) Panasonic high-definition plasma television (model number TC-P50S2).

Table 1 Model demographics, maximum vertical jump (RWJ) heights, and horizontal jump (LJ) distances
Full size table

Procedure

Participants sat in a chair approximately 127 cm away from the television screen. The television sat on a table and was adjusted in height for each participant such that the screen center closely aligned with participants’ sitting eye heights. Point-light videos appeared in the lower left portion of the screen. Participants were instructed to watch each video, then use a wireless mouse situated on the armrest of the chair to move a small star on the screen to the maximum height they believed the model could reach with their fingertips while jumping vertically from a standing position (for RWJ reports; Fig. 1, left) or the maximum distance they believed the model could cover (i.e., where his/her heels would land) while jumping horizontally from a standing position at the red box (for LJ reports; Fig. 1, right). The initial screen position of the star was randomized, and its final location could be adjusted by participants before moving to the next trial. Perceptual reports were provided twice for each of the 12 point-light videos, yielding a total of 24 trials for each affordance type. Trial order was randomized within each affordance type and was counterbalanced across participants.

Fig. 1
figure 1

Screen display for RWJ reports (left) and LJ reports (right)

Full size image

Results and discussion

Raw perceptual reports (e.g., the screen location of the star’s final position) were converted into metric units (centimeters) such that reports for RWJ represent vertical distance from the floor and reports for LJ represent horizontal distance from the red start box. Accuracy of perceptual reports was evaluated using the ratio of the average report provided by each participant divided by the actual affordance boundary of each respective model. Ratios of 1 indicate perfect accuracy, ratios greater than 1 indicate an overestimation in affordance reports, and ratios less than 1 indicate an underestimation. Mean ratios were submitted to a 3 (perceiver type: basketball player vs. soccer player vs. nonplayer) × 4 (video type: balance vs. twist vs. squat vs. walk) × 3 (model: 1 vs. 2 vs. 3) × 2 (affordance type: RWJ vs. LJ) mixed ANOVA with perceiver type as a between-subjects factor and video type, model, and affordance type as within-subjects factors.

Role of movement type (related/unrelated to action)

A main effect of video type was found, F(3, 72) = 18.69, p < .0001, 1 − β = 1.00, η p 2 = .44. Bonferroni-corrected post hoc tests revealed that reports for balance and twist were significantly different from reports for squat (ps < .0001), as well as reports for walk (ps < .0001), but no differences were found between reports for balance and twist (p = .25) or between reports for squat and walk (p = .64). Ratios for squat and walk were closer to 1 than ratios for balance and twist, confirming our prediction that perceptual reports would be more accurate after movements related to performing a vertical and a horizontal jump were viewed (Fig. 2, left). These results support previous findings that perception of action-scaled affordances is more accurate after movements related to the affordance are observed (Ramenzoni et al., 2010; Ramenzoni, Riley, Davis, et al., 2008; Weast et al., 2011) and indicate that point-light displays depicting movements related to jumping appear to capture the kinematic information that is essential to (i.e., necessary for) the accurate perception of RWJ and LJ.

Fig. 2
figure 2

Mean ratios of provided affordance reports to actual model affordance boundaries are displayed for main effects of video type (left) and perceiver type (center), as well as the interaction of model and affordance type (right). The dashed line indicates perfect accuracy (a ratio of 1)

Full size image

Influence of sports experience

A main effect of perceiver type was found, F(2, 24) = 4.68, p = .02, 1 − β = 0.73, η p 2 = .29. Bonferroni-corrected post hoc tests revealed that reports provided by nonplayers were significantly different from reports provided by basketball players (p = .01) and from reports provided by soccer players (p = .02); however, there was no difference between reports from basketball and soccer players (p = .93). Both basketball players and soccer players exhibited ratios closer to 1 than ratios from nonplayers, indicating that both groups of athletes were more accurate than nonplayers in perceiving both types of affordances, regardless of relevance to one’s sport (Fig. 2, center,). This may indicate that athletes have become attuned to information specifying affordances generally, rather than only for those relevant to their sport. It is also possible, though, that the choice of athletes was not appropriate for distinguishing differences in perception of relevant versus nonrelevant affordances. For example, we hypothesized that basketball players would perceive the RWJ affordance more accurately, given their presumed greater experience, but it might be the case that soccer players have sufficient experience in observing others reaching while jumping to perform at the same level of accuracy as basketball players. Future research must include athletes in sports that do not involve ratcheting others jump (e.g., long-distance running) to determine whether perception is more accurate for affordances related to one’s sport.

Differences in accuracy for RWJ and LJ

Unexpectedly, main effects were found for both affordance type, F(1, 24) = 14.54, p = .0008, 1 –−β = .97, η p 2 = .38, and model, F(2, 48) = 55.65, p < .0001, 1 − β = 1.00, η p 2 = .70. However, these factors also interacted, F (2, 48) = 32.74, p < .0001, 1 − β = 1.00, η p 2 = .58. Simple-effects analyses revealed a simple main effect of model for LJ reports, F(2, 24) = 77.27, p < .0001, 1 − β = 1.00, η p 2 = .76, but not for RWJ reports (p = .10, 1 − β = .46, η p 2 = .09). Bonferroni-corrected post hoc analyses revealed that LJ reports were significantly different for each model (all ps < .0001); perceivers were increasingly more accurate for models 1–3, respectively (Fig. 2, right). Participants were least accurate at differentiating the models’ LJ capabilities. No other interactions were significant (all ps > .05).

It is possible that biological motion perception is influenced by an observer’s familiarity with the observed action (Calvo-Merino, Grezes, Glaser, Passingham, & Haggard, 2006; Casile & Giese, 2006; Hohmann, Troje, Olmos, & Munzert, 2011). Interestingly, Casile and Giese found that motor experience in performing an action improves visual action perception, even when vision is absent from the motor-learning experience. When visual recognition of novel gait patterns was assessed both before and after blindfolded motor training, observers improved in visually recognizing only the gait pattern for which they had received motor training. A possible reason why reports were more accurate for RWJ than for LJ may be because participants have less motor experience performing LJ than performing RWJ, preventing better perception of this action boundary for others.

Observers demonstrated high degrees of accuracy in perceiving RWJ when provided kinematic information in walking (Weast et al., 2011) and in squatting (Ramenzoni et al., 2010). It is possible, though, that these two movement types do not contain sufficient information for perceiving another’s ability to perform a long jump. Standing long and vertical jumps employ different coordination strategies, each utilizing muscle groups differently during these “explosive” actions; hip muscles contribute the majority of mechanical work done during the propulsion phase of a vertical jump, whereas ankle muscles provide the greatest contribution for a long-jump (Nagano, Komura, & Fukashiro, 2007; Robertson & Fleming, 1987). Walking, therefore, may not provide information about the force-production capabilities of the ankles (since hip muscles contribute the majority of mechanical work during walking; Neptune, McGowan, & Kautz, 2009), nor may squatting (for which there is little movement at the ankles), preventing accurate perception of another’s maximum LJ distance. Further investigation is required to determine the reason for poorer performance in predicting LJ, possibly by including additional stimulus videos of movements better related to LJ (e.g., calf-raises).

This research establishes that point-light displays contain essential kinematic information for action-scaled affordance perception; observers could accurately perceive the action boundaries of other actors after viewing point-light motion depicting only the actors’ joint centers. Additionally, we found further evidence that long-term experience in a sport results in attunement to kinematic information specifying domain-relevant affordances for others: Both groups of athletes were more accurate than nonplayers in perceiving affordances for the point-light actors. Although our findings support previous claims that viewing affordance-related movements improve affordance perception, future research must evaluate whether this improvement occurs only after viewing movements that recruit the same primary muscle groups as those recruited for the affordance in question. Lastly, in order to determine the information to which athletes are sensitive, future research must focus on identifying the structure in movement kinematics that specifies action-specific affordances.