Edited by: Qasim Zaidi, SUNY College of Optometry, United States
Reviewed by: Erin Koch, California Institute of Technology, United States; Kevin R. Brooks, Macquarie University, Australia
This article was submitted to Perception Science, a section of the journal Frontiers in Psychology
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
The proficiency of human observers to identify body postures is examined in three experiments. We use a posture decision task in which participants are primed with either anatomically possible or impossible postures (in the latter case the upper and lower body face in opposite directions). In a long-term priming paradigm (i.e., in an initial priming block of trials and a subsequent test phase several minutes later), we manipulate the relation between priming and test postures with respect to the identity of the person in the body postures (Experiment 1), the prototypicality of the depth orientations (Experiment 2), and the variability of the priming orientations (Experiment 3). Reaction time to the test postures is the main dependent variable. In Experiment 1 it is found that priming of postures does not depend on the exact visual appearance of the actor (either same priming and test female or male figure or different figures), supporting the hypothesis that posture priming primarily is determined by the spatial relations between the body parts and much less by characteristics of the person involved. Long-term priming in our paradigm apparently is based on the reactivation of high-level posture representations that make abstraction of the identity of the human figure. In Experiment 2 we observe that privileged or prototypical orientations (e.g., 3/4 views) do not affect long-term priming of body postures. In Experiment 3, we find that increasing or decreasing the variability between the priming and test figures influences reaction time performance. Collectively, these results provide a better understanding of the flexibility (e.g., invariant to identity) and limits (e.g., depending on depth orientation) of the processes supporting human posture recognition.
Human observers exhibit an impressive level of proficiency in identifying the body postures of conspecifics (e.g.,
A snapshot of a running action seen from 4 different viewpoints.
In order to investigate the nature of the representations underlying visual perception of human body postures,
Examples of anatomically possible (first column) and impossible (second column) postures.
Illustration of the basics of the long-term priming paradigm.
Our underlying working hypothesis is that priming results from the persistent activation of representations that mediate perceptual organization of the posture. In order to perform the posture decision task during the priming phase, participants compute a representation of the posture and, when they re-encounter the posture during the test phase, activation of the representation is facilitated, resulting in shorter reaction times. The finding that even a relatively moderate depth rotation of the body posture between priming and test phase already results in a drastic reduction of facilitatory priming suggests that the underlying representations that mediate visual posture identification are sharply tuned to specific depth orientations.
However, there is an alternative explanation. Facilitatory priming was strongly dependent on the repetition of exactly the same priming posture in all its details. Therefore, it is possible that priming was based on early, lower-level stimulus-specific representations that are only precursors to higher-level body representations. The observation that there was no facilitatory priming for impossible postures (see
Long-term priming was examined with the paradigm developed by
First, and most importantly for the present experiment, we manipulated the visual appearance of the actor involved in the postures. As shown in
The two actors (one male, one female) shown in Experiment 1.
Illustration of the priming conditions in Experiment 1. Top panel: same orientation/different figure condition; bottom panel: different orientation/same figure condition. First row in first panel: priming condition (same or different figure and same orientation, or no prime). First row in second panel: priming condition (same figure and same or different orientation, or no prime). Second row in both panels: primed condition (here figure in 45° orientation).
The second purpose of Experiment 1 was an attempt to replicate the strong viewpoint dependence observed by
For a given posture there were four different conditions: same figure/same orientation in priming and test block, different figure/same orientation, same figure/different orientation, and no prime. In principle, both the visual appearance of the human model and the precise depth viewpoint from which a posture is observed are irrelevant to perform the posture decision task. The observation that long-term priming with same figure/same orientation primes generalizes over changes in characteristics of the model but not over changes in the depth orientation of the posture would support the hypothesis that the high-level representations of human postures are viewpoint specific. Moreover, the observation that same-orientation prime-test pairs produce (relatively) equivalent facilitatory effects independently of the precise visual appearance of the actor (either same priming and test female or male figure or different figures) would indicate that posture priming primarily is determined by the spatial relations between the body parts and much less by characteristics of the person involved in the posture. This would imply that high-level posture representations to a large extent make abstraction of the identity of the human figure.
Sixty-four first-year students (who received course credit) or other undergraduate and graduate students (who were paid) participated in the experiment. They were tested individually. The subjects gave informed consent in accordance with the declaration of Helsinki. All participants had normal or corrected-to-normal vision and were naive with respect to the hypothesis under investigation.
The experimental stimulus set consisted of eight sets of 24 static full-color pictures of anatomically possible human body postures. In the first set, a male bare-footed model with short brown hair and a blue trouser suit with short sleeves and trouser-legs was depicted in 24 different postures with the trunk in a 15° orientation in depth (with 0° corresponding to a frontal view). In the second set, the same model was shown in the same 24 postures, but now in a 45° depth orientation. The third and the fourth set portrayed the human figure in 24 new postures, with the trunk in a 195 and a 225° depth orientation, respectively. In the four remaining sets, the same postures were shown as in the previous four sets, but the model now was a female person with medium blond hair wearing green shoes and a green trouser suit with long sleeves and legs. Care was taken that as many body parts as possible were visible in all views.
In addition to the experimental stimulus set, 60 filler stimuli of 40 anatomically impossible postures were constructed. In these impossible poses, the above-waist part of the body was rotated 180° in depth around the model’s top-bottom axis, so that the figure’s above-waist part was oriented in exactly the opposite direction vis-a-vis the below-waist part of the body (
The training stimuli consisted of the 14 training stimuli used in Experiments 3 and 4 of
All anatomically possible and impossible postures were created using the Poser software package (
Each trial started with the presentation of an auditory warning signal and a fixation cross in the middle of the screen. After 500 ms, the stimulus appeared and participants had to decide as rapidly as possible whether the depicted body posture was anatomically possible or not, by pressing one of two response keys. Half of the participants pressed the right button for possible poses and the left button for impossible poses, while this stimulus-response mapping was reversed for the other half of the subjects. Auditory feedback was given by means of a high-pitch tone for a correct answer and a low-pitch tone for an incorrect answer. The picture was presented until participants responded, except when the RT exceeded 2 s, in which case the trial was ended.
Participants performed this posture decision task with the 14 training stimuli, immediately followed by the priming block of 68 priming stimuli (36 experimental, anatomically possible postures and 32 filler, impossible postures) presented in an individually determined random order. After a 5 min break (during which the experimenter and the subject had an informal conversation), the testing block of 88 stimuli (48 experimental, anatomically possible postures and 40 filler, impossible postures) was administered, again in a random presentation order. During instructions, it was never mentioned that subjects were involved in a priming experiment.
There were four conditions (
All participants saw the same anatomically impossible filler stimuli. During the priming block and the test block, 32 and 40 impossible postures were shown, respectively, half of them personated by the male model and half by the female model. In both phases, half of the male and half of the female postures were oriented (more or less) toward the viewer (15 or 45° in the priming phase and 45° in the test phase) and half were oriented (more or less) away from the viewer (195 or 225° in the priming phase and 225° in the test phase). Twelve impossible postures were shown by the same figure and in the same depth orientation in priming and test phase, 10 impossible postures were performed by the same figure but shown from a different viewpoint in the test phase, and 10 impossible postures were shown from the same viewpoint, but were performed by the other person in the test phase. The remaining eight impossible filler stimuli were only administered during the test phase.
The dependent variable was the RT to the (anatomically possible) test postures. Trials in which the stimulus was not identified correctly either in the priming or in the test block and trials in which the RT fell below a cut-off value of 200 ms or above a cut-off value of 2000 ms were discarded from the RT data set (approximately 1% of the data set). The remaining RTs were entered in a subject repeated-measures analysis of variance (ANOVA) with priming condition (identical prime, different figure but same orientation, same figure but different orientation, and no prime) as a within-subject variable and participant group as a between-subjects variable, and in a stimulus ANOVA with priming condition as within-stimulus variable and stimulus group as between-stimuli variable. (Especially in psycholinguistic research, but also in perception research, it is informative to perform both subject and stimulus analyses and present them together,
Mean identification time (in ms) of anatomically possible human postures in the test phase of Experiment 1 as a function of long-term priming condition.
Long-term priming condition | RT to possible body postures |
Same figure/same orientation | 615 |
Different figure/same orientation | 624 |
Same figure/30° different orientation | 621 |
No priming | 646 |
Both the subject and the stimulus analysis revealed a statistically significant main effect of priming condition (4 levels: same figure, same orientation; different figure, same orientation; same figure, different orientation, no prime),
Note that the long-term priming effects in the test phase were not caused by (accidental) differences in exposure times to the initial, priming postures. Indeed, a subject and stimulus analysis on the RTs in the priming phase showed that long-term priming condition had no effect,
First, in comparison to the no-prime baseline, participants were faster to decide that a posture was anatomically possible when they had seen that posture several minutes earlier during the priming phase. Most importantly, this priming effect was not significantly larger when the human model in the priming posture was identical to the model in the test phase than when priming and test postures were personated by distinctly different human models. Apparently, facilitatory long-term priming in the posture decision task is not contingent upon the repetition of exactly the same stimulus person. This supports the hypothesis that long-term priming is based on the re-activation of high-level representations of human body postures (rather than being based on an early, low-level representation of the stimulus) that make abstraction of the precise visual appearance of the human figure.
Second, contrary to our expectations, priming also generalized over an orientation difference of 30° between priming and test posture. On the one hand, given the fact that the difference between postures in a 45 and a 15° orientation and between a 225 and 195° orientation was quite subtle (see
A possible explanation for the contradictory findings in our previous experiments hinges on the hypothesis that some views of body postures might have a privileged status over other views and that posture recognition proceeds through the activation of these privileged or prototypical orientations. Evidence mainly comes from studies of object recognition.
First, there is ample evidence that some views of three-dimensional, familiar objects are rated as more canonical or prototypical than other views and that objects depicted in canonical orientations are identified more easily (e.g., faster) than when shown from less canonical angles (e.g.,
Second, it has been suggested that identification of objects, even objects viewed from unconventional viewing angles, is achieved by activating (a number of) neighboring prototypical views (for an overview on this discussion, see
The test postures in Experiment 4 of
In Experiment 2, the test postures were shown either in a prototypical 45 or 225° orientation (further referred to as the three-quarter views) or in a less prototypical 75 or 255° orientation (further referred to as the sagittal views, even though strictly spoken the views only approximate the 90 and 270° sagittal views; note that one of the reasons for choosing these views close to the sagittal views instead of the exact sagittal views is that body parts that were occluded in the sagittal views mostly became visible in the close-to-sagittal views). These test postures were preceded by a posture in the same view, the same posture in a view that differed by a rotation of 30° (15 or 195° and 45 or 225°), or were not shown during the priming phase. Of crucial importance is the condition in which priming and test postures differed by 30°. As spelled out in previous paragraphs, facilitatory priming (i.e., shorter RTs in comparison to the no-prime condition and RTs at the same level as in the identical prime condition) was predicted in this condition for test postures in prototypical three-quarter orientations but not for test postures in less prototypical sagittal orientations. The underlying rationale is that the 30° different prime preceding the three-quarter test posture is a less prototypical view and processing this view also activates the three-quarter view, whereas the 30° different prime preceding the sagittal test posture is a prototypical view and processing this view does not lead to activation of less prototypical views. Support for this hypothesis would imply that, first, postures in some orientations have a more privileged status than postures in other orientations and, second, posture decision proceeds through the activation of neighboring privileged views.
A total of 84 first-year students psychology at the Leuven University with normal or corrected-to-normal vision participated for course credit. Participants were tested individually.
The experimental stimulus set consisted of six sets of 24 color pictures of a male figure. The first three sets depicted the figure in 24 different anatomically possible body postures with the trunk in a 15, 45, or 75° depth orientation, respectively. In the other three sets, the same figure was shown in 24 other postures in a 195, 225, or 275° depth orientation, respectively.
Example of the stimuli used in Experiment 2. The trunk is in a 15°, 45°, or 75° degrees depth orientation (top row) or a 195°, 225°, or 275° depth orientation (bottom row).
Participants performed the same posture decision task as in Experiment 1. All participants saw exactly the same test postures, but different subjects saw different priming postures.
The six sets of experimental stimuli were divided in six groups of four stimuli that were rotated across conditions and across participants. Each participant was presented with eight different postures in each of six conditions. In three conditions, three-quarter views (45 and 225°) were shown during the test phase and in three other conditions, views close to the sagittal orientation (75 and 255°) were shown during the test phase. One third of the three-quarter and quasi sagittal test postures were preceded by the same posture in the same orientation during the priming phase, one third of the test postures were preceded by the same posture rotated by 30° in depth vis-a-vis the test stimulus (i.e., in a 15 and 195° orientation for the three-quarter views and a 45 and 225° orientation for the sagittal views), and one third of the test postures were not shown during the priming phase. All participants saw the same anatomically impossible postures. In half of the impossible postures presented during the test phase, the lower body part half was in a 45 or 255° depth orientation, in the other half, the lower body part was in a 75 or 255° orientation. One third of the impossible test postures were seen in the same depth orientation during the priming phase, one third in a 30°-different orientation, and one third was seen for the first time during the test block of trials. These impossible postures only served as filler stimuli and RTs to impossible postures were not analyzed.
Participants were tested individually. They started the experiment with a block of 14 training stimuli (the same as in Experiment 1), which was followed by a priming block of 56 stimuli presented in a random order (32 possible experimental stimuli and 24 impossible filler stimuli), a 5 min break, and a test block of 84 stimuli in a random order (48 experimental possible stimuli and 36 impossible filler stimuli).
The RT to decide that a test posture was anatomically possible was the dependent variable. Using the same criteria as in Experiment 1, about 1% of the trials were excluded from analysis. Means of the remaining RTs, as a function of priming condition and test orientation are shown in
Mean identification time (in ms) of anatomically possible human postures in the test phase of Experiment 2 as a function of long-term priming condition and test orientation.
Test orientation |
||
Long-term priming condition | Three-quarter view | Sagittal view |
Same orientation | 616 | 616 |
30° different orientation | 638 | 625 |
No priming | 658 | 650 |
The data were entered in a participant ANOVA with test orientation (three-quarter view or almost sagittal view) and priming condition (same orientation prime, 30° orientation-different prime, or no prime) as within-subject variables and participant group as between-subjects variable, and in a stimulus ANOVA with test orientation and priming condition as within-stimulus variables and stimulus group as between-stimuli variable. The participant and the stimulus analysis yielded a significant main effect of priming condition,
The main effect of test orientation was not significant,
Note that again the priming effects in the test phase were not caused by differences in initial identification time during the priming phase, as shown by the absence of main effects of test orientation and priming condition and the absence of an interaction effect in a participant and stimulus analysis on the reaction times in the priming phase. With the three-quarter view test orientation, mean identification time of the priming posture was 862 ms for the same-orientation condition and 852 ms for the 30° different-orientation condition. With the almost frontal view test orientation, mean identification time of the priming posture was 851 ms for the same-orientation condition and 854 ms for the 30° different-orientation condition.
For both test orientations in Experiment 2, we observed long-term priming that was less orientation specific (i.e., priming in the 30° difference condition larger than in the no-prime baseline but smaller than in the identical orientation condition) than in Experiment 4 of
From a theoretical point of view, these findings have implications for a better understanding of visual representations of human body postures. More specifically, we did not find evidence for the assumption that postures in a three quarter view are processed as prototypical postures nor that body postures in nearby orientations are recognized via the activation of prototypical half-way orientations. Note, however, that the hypothesis that prototypical orientations play a role in the identification of postures in less prototypical orientations can only be rejected with caution. The assumption underlying this hypothesis was that the presence of non-prototypical orientations in the priming phase could result in the activation of similar prototypical views, so that these prototypical orientations are processed more easily later during the test phase. However, this facilitatory effect in principle could also occur when prototypical orientations are shown during the priming phase and non-prototypical orientations during the test phase. In theory it is indeed possible that the presentation of prototypical orientations in the priming phase results in faster activation of these representations in the test phase, facilitating the activation of test postures in non-prototypical orientations that are recognized through the activation of these representations. What does seem to be clear on the basis of Experiment 2, however, is that prototypical orientations are not responsible for the divergent results in Experiment 4 of
The design of Experiment 4 in
Under the assumption that different body postures in orientations that are closer to each other are more similar than postures in orientations that are farther apart, this fits with findings in the object recognition literature. It has indeed been shown that object identification becomes more orientation specific as objects in the stimulus set become more similar (e.g.,
Examples of postures with the trunk in a 45° (first row) or 225° (second row) depth orientation. Note that the orientation of the trunk more or less remains the same in each row.
The orientation differences in Experiment 4 of
Illustration of the fact that small rotations result in less drastic changes in the projection of the trunk than large rotations (first row: two postures in more or less frontal view (75 and 60°) that differ by 15° in depth; second row: two postures in more or less frontal view (75 and 45°) that differ by 30° in depth; third row: two postures in more or less back view (255 and 240°) that differ by 15° in depth; fourth row: two postures in more or less back view (255 and 225°) that differ by 30° in depth).
It has been suggested (e.g.,
The purpose of Experiment 3 was to examine to what degree the extent of the orientation differences between different body postures in the priming phase could explain the divergent results of Experiment 4 of
Two groups of 60 subjects participated to the experiment. All participants had normal or corrected-to-normal vision and were tested individually. Subjects either were first or second year students who participated for course credit or were phd students.
A total of 16 stimulus sets from Experiment 4 in
Examples of stimuli used in Experiment 3 (first row: possible posture in 15, 45, and 75° depth orientation; second row: possible posture in 195, 225, and 255° orientation; third row: impossible posture with lower body in 15, 45, and 75° orientation; fourth row: impossible posture with lower body in 195, 225, and 255°).
Participants were randomly assigned to the two groups. The first group participated in the condition with large orientation differences between the stimuli in the priming phase and the second group participated in the condition with small orientation differences in the priming phase. The stimuli in the test phase always were oriented 75 or 255° to the right. In the condition with large orientation differences stimuli in the priming phase either had the same orientation as in the test phase (75 or 255°), a 30° different orientation (45 or 225°), a 60° different orientation (15 or 195°), or there was no priming. In the condition with small orientation differences stimuli in the priming phase had the same orientation as in the test phase (75 or 255°), a 15 difference (60 or 240°), a 30° difference orientation (45 or 225°), or there was no priming. The smallest orientation difference between different stimuli in the priming phase therefore was 30° in the condition with large orientation differences and 15° in the condition with small orientation differences. In both conditions 12 anatomically possible and 12 impossible stimulus sets were divided over 4 groups of 6 stimuli that were rotated across long-term priming conditions and participants. Each participant was presented with 12 anatomically possible and 12 impossible postures in each long-term priming condition. Half of these 12 anatomically possible and 12 impossible postures were shown in a (more or less) frontal view (15, 45, and 75 or 45, 60, and 75°) and the other postures in a (more or less) back view (195, 225, and 255 or 225, 240, and 255°). The experiment always started with the training stimuli and 72 priming stimuli presented in a random order. After a short break, the 96 test stimuli were administered, also in a random order.
Reaction times below 200 ms or above 1400 ms (about 1% of the data), reaction times to stimuli that were not identified correctly in the priming phase or the test phase, and reaction times to three impossible postures that were identified correctly by less than half of the participants in both phases were removed from the data set. Mean RTs to anatomically possible and impossible postures can be found in
Mean identification time (in ms) of possible and impossible human postures in the test phase of Experiment 3 as a function of context condition (large vs. small orientation differences) and long-term priming condition (same orientation, 15, 30, or 60° difference, or no priming).
Possible postures |
Impossible postures |
|||
Priming condition | Large orientation difference | Small orientation difference | Large orientation difference | Small orientation difference |
Same ori | 640 | 628 | 709 | 728 |
15° difference | – | 622 | – | 714 |
30° difference | 642 | 644 | 710 | 718 |
60° difference | 660 | – | 720 | – |
No priming | 662 | 650 | 721 | 743 |
In a first series of analyses, RTs to stimuli in the 60° difference condition in the first group of participants (who were presented with relatively large orientation differences, but not the 15° difference) and RTs to stimuli in the 15° difference condition in the second group (who were presented with relatively small orientation differences, but not the 60° difference) were not taken into account. All other RTs from the test phase were entered in a participant and stimulus repeated-measures ANOVA with range of orientations (relatively large or small orientation differences in the priming phase) as between subjects variable or within stimuli variable, long-term priming condition (same orientation condition, 30° difference condition, or no priming) as within subjects variable or within stimuli variable, stimulus type (anatomically possible or impossible postures) as within subjects variable or between stimuli variable, and subject group or stimulus group as between subjects variable or between stimuli variable.
In the subject analysis as well as in the stimulus analysis, there was a main effect of stimulus type,
Even though stimulus type only interacted with range of orientations condition and not with long-term priming, separate analyses were performed for the anatomically possible and impossible postures. Differences between the same orientation condition, the 30° difference condition, and the condition without priming were evaluated by means of Dunn’s multiple comparison procedure. In the possible posture condition with large orientation differences, the difference between RTs in the same orientation condition and the condition without priming and in the condition with 30° different orientations and the condition without priming were significant in the subject and the stimulus analysis,
The data indicate that long-term priming effects are modulated by the magnitude of the orientation differences between the stimuli in the priming phase. This is supported when also reaction times to stimuli in the 60° difference condition in the group of participants who were presented with large orientation differences and the reaction times to stimuli in the 15° difference condition in the group of participants who were presented with small orientation differences were taken into account (
In the condition with impossible postures (
A subject and stimulus analysis on the RTs in the priming phase (
Mean identification time (in ms) of possible and impossible human postures in the priming phase of Experiment 3 as a function of context condition (large vs. small orientation differences) and long-term priming condition (same orientation, 15°, 30°, or 60° difference).
Possible postures |
Impossible postures |
|||
Priming condition | Large orientation difference | Small orientation difference | Large orientation difference | Small orientation difference |
Same orientation | 919 | 889 | 944 | 968 |
15° difference | – | 912 | – | 956 |
30° difference | 929 | 918 | 950 | 991 |
60° difference | 939 | – | 1019 | – |
In an analysis of the mean identification times to anatomically possible postures in the priming phase as a function of long-term priming condition and range of orientations condition there was only one reliable difference in the subject analysis. Participants were faster to respond in the same-orientation condition (75 and 255° orientations) than in the 30° difference orientation condition (25° and 225° orientations) in the condition with small orientation differences,
The analysis of the mean identification times for the impossible postures in the priming phase as a function of range of orientations condition and long-term priming condition indicated that in the condition with large orientation differences participants reacted more slowly in the 60° difference condition (15 and 195° orientation) than in the two other conditions (
We performed additional analyses on RTs for anatomically possible poses and for impossible poses in the priming phase with long-term priming condition, range of orientations condition, and global orientation (frontal or back view) as independent variables, but there were no significant interaction effects.
Experiment 3 showed that long-term priming is influenced by the extent of the orientation differences between the stimuli in the priming phase. When the orientation difference between stimuli was at least 30°, there were significant long-term priming effects in the 30° different-orientation condition. When a number of stimuli only differed by 15°, no long-term priming was observed in the same 30° different-orientation condition. These results can be interpreted in at least two ways. First, it is possible that the presence of a large 60° orientation difference resulted in broader tuning of the representational system, leading to more generalization over orientations. However, this explanation is improbable. The presence of a 60° different orientation condition in Experiment 4 of
Apparently, not the presence of a 60° difference in the condition with large orientation differences, but the presence of a 15° difference in the condition with small orientation differences was crucial. Small orientation differences in the priming phase of a long-term priming experiment seem to result in finer orientation tuning of the representations that are used to identify body postures. This is remarkable. As can be observed in
It has been shown before that the extent of generalization for a stimulus in a particular orientation is variable and that it depends on the circumstances under which identification takes place. For example, when two orientations are connected by apparent motion, representations are tuned in such a way that short-term priming between the two orientations is facilitated, whereas priming outside the movement path is inhibited (
This observation fits with findings on object perception. Indeed, it has been reported repeatedly that object recognition becomes more orientation specific as the similarity of the objects in the stimulus set is more pronounced (
In
Common representational space of two human body postures (each column depicts a particular postures in four different depth orientations; in each row two similar but distinct postures in the same depth orientation).
The model in
The context (i.e., the range of orientations used) dependent long-term priming effects in Experiment 3 suggest that the underlying representational system is flexible and that posture identification is optimized by adapting the degree of overlap between the representations to the stimulus context (or more generally, the task at hand). A stimulus set with large orientation differences between different stimuli gives rise to more broadly tuned representations than a stimulus set with small orientation differences. Apparently, representations are more finely tuned when identification involves making finer discriminations. Representations that are smaller tuned are closer to the original stimulus and cause less generalization. By using more narrowly tuned representations, it is avoided that similar stimuli in similar orientations lead to the activation of strongly overlapping representations that make precise identification difficult.
Tuning of representations or generalization fields can be conceptualized as an adaptation (during the task) of the manner in which neurons and neuron populations are activated. When, as the task proceeds, it becomes obvious that the normal activation spreading of neuronal populations results in a larger overlap for different stimuli, a stronger criterion is set, as a result of which smaller tuning is established. In a long-term-priming experiment this adaptation primarily occurs during the priming phase. When confronted with a stimulus in the priming phase groups of neurons are tuned as a function of a number of stimulus characteristics. When during the test phase the same stimulus with the same characteristics is shown again, the identification threshold is reached faster. If, however, there is a significant change in a stimulus characteristic, partly different neurons are activated, resulting in no or less facilitation. The stimulus context (e.g., range of orientations used) probably codetermines the breadth of tuning of the neurons or the extent of activation spreading over neurons and neuron populations.
How do observers identify human body postures? Is it the case that for all possible orientations of a body postures a single orientation-independent, object-centered memory model can be accessed in order to identify the posture (e.g.,
We were not able to directly address the inconsistency between the present Experiment 1 (i.e., priming across a 30° difference between and test posture) and research previously described by
As already indicated, the findings in the long-term priming experiments are in line with observations on object recognition. On the one hand, this suggests that the visual system uses similar mechanisms and representations to identify objects and postures in different orientations. On the other hand, identification of body postures seems to be special. When observers are confronted with a human figure, it is probable that a general body scheme is activated that is adjusted in the correct posture and orientation. In the experiments reported in
Neurophysiological studies confirm the existence of a body-specific representational system (e.g.,
There is evidence that region MT/V5 is strongly associated with motion perception (e.g.,
Some action sequences consist of changes in the global orientation of the human figure who performs the action (e.g., rotating movement as in a pirouette). In this case integration of different action phases boils down to integration of different orientations. In short-term priming experiments, it has been shown before (e.g.,
In sum, the dynamic orientation-specific representations supporting posture perception are flexible and dependent on stimulus and task context. This allows the visual system to achieve a broad range of tasks. Successful identification of highly similar body postures in similar orientations probably is best supported by more finely tuned representations, whereas anticipation of future orientations and action phases (e.g.,
The datasets generated for this study are available on request to the corresponding author.
The studies involving human participants were reviewed and approved by the Ethics Committee of the Faculty of Psychology and Educational Sciences. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.
Both authors contributed to framing the study theoretically, designing and executing the experiments, analyzing the data, interpreting the results, and drawing conclusion.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
We are grateful to Filip Germeys for his comments on this research project.
Note that the link between low and high-level processing on the one hand and early and later processing mode is a hypothesis and not every researcher agrees with the link (see