The Psychophysics of Chasing
Here are some demonstrations of the various conditions discussed in the following paper:
Gao, T., Newman, G. E., & Scholl, B. J. (2009). The psychophysics of chasing: A case study in the perception of animacy. Cognitive Psychology, 59(2), 154-179.
These demonstrations are provided as Quicktime movies, which can be downloaded or viewed directly in most web-browsers. These movies are a bit large and choppy, but they should be sufficient to illustrate the basic conditions. As highly compressed versions of the original stimuli, these movies may not preserve the precise spatial and temporal characteristics of the originals.
The currency of our visual experience consists not only of visual features such as color and shape, but also seemingly higher-level features such as animacy. Psychologists have been captivated by the fact that even simple moving geometric shapes may appear animate and goal-directed, but previous studies have faced two major challenges: (a) It is difficult to measure perceived animacy with quantitative precision. (b) Task demands make it difficult to distinguish perception from higher-level inferences. In our paper, we introduce two methods that avoid these concerns, in the context of what is perhaps the simplest example of a perceived social interaction: one shape (the wolf) chasing another shape (the sheep) in a display containing several featurally-identical items. We then use these methods to explore two new cues to animacy: chasing subtlety and directionality.
Experiment 1: The Search for Chasing
On each trial in the Search-for-Chasing task, subjects simply had to report whether or not the animation contained a chase -- with a wolf pursuing a sheep. Across trials, we manipulated chasing subtlety -- the degree to which the wolf's approach could deviate from perfectly 'heat-seeking' pursuit. We provide each animation below in two versions. One approximates the displays actually seen by the subjects, in which all objects were featurally identical. In the other 'Cheat' version, the wolf is drawn in red, and the sheep is drawn in green. This may help to orient readers to the subtler conditions, but note that our subjects only ever saw monochromatic displays. Note that the animations below depict only 'chase-present' search trials. On chase-absent trials, as detailed in our paper, the wolf was still present, but was 'chasing' an invisible sheep. The results (1) demonstrate how the perception of animacy can be measured with precision; (2) demonstrate how we can evaluate the objective accuracy of chasing detection; and (3) help distinguish the immediate perception of chasing from those more subtle (but nevertheless real) types of 'stalking' that cannot be readily perceived.
Experiment 2: The Search for Correlations with 'Phantom Sheep'
Baseline animation (Chasing subtlety = 180 deg) (864 KB)
Baseline animation (Chasing subtlety = 180 deg) [Cheat] (1.1 MB)
This is a baseline condition, in which there is no statistical correlation between the wolf's trajectory and the sheep's movement -- and thus no actual chasing at all.
Perfect heat-seeking (Chasing subtlety = 0 deg) (808 KB)
Perfect heat-seeking (Chasing subtlety = 0 deg) [Cheat] (1 MB)
Here the wolf 'heat-seeks' the (moving) sheep -- with the wolf updating its position every 170 ms so as to be heading directly toward the sheep.
Mild subtlety (Chasing subtlety = 30 deg) (860 KB)
Mild subtlety (Chasing subtlety = 30 deg) [Cheat] (1 MB)
Here the wolf is always heading in the general direction of the sheep, but is not perfectly heat-seeking: instead, the wolf's trajectory is determined randomly with the contstraint that each update keeps it heading within 30 deg of the sheep's true direction (for a total angular window of 60 deg, centered on the line connecting the wolf and sheep).
Large subtlety (Chasing subtlety = 90 deg) (860 KB)
Large subtlety (Chasing subtlety = 90 deg) [Cheat] (1 MB)
Here the direction in which the wolf moves is even less constrained, to within 90 deg of the sheep's true direction (for a total angular window of 180 deg, centered on the line connecting th wolf and sheep). Thus the wolf may now head (in the most extreme case) in an orthogonal direction to the (moving) sheep, but can never be heading away from it.
To demonstrate that the results of Experiment 1 reflect the detection of chasing per se -- and not just any such correlations between movements -- this experiment contrasted two equally strong types of motion correlations. Chasing trials were identical to those of Experiment 1. On Mirror trials, though, the wolf was chasing a 'phantom sheep' -- the moving point that was the reflection of the actual sheep's location through the center of the display. (Note that as a result, the wolf still was frequently still heading toward the actual sheep; see the text for details.) As a result, the degree of statistical correlation between the movements of the wolf and sheep were identical in each condition. Correlation detection was much better on Chasing trials than on Mirror trials, demonstrating that the perception of chasing can be dissociated from more general forms of spatiotemporal correlations.
Experiment 4: Don't Get Caught!
Correlation Control (Chasing subtlety = 30 deg) (868 KB)
Correlation Control (Chasing subtlety = 30 deg) [Cheat] (1.2 MB)
These trials depict a Target Present trial of the Mirror condition, with a chasing subtlety of 30 deg. The first movie shows what an actual trial looked like, with monochromatic objects. In the 'Cheat' version, (1) the wolf is drawn in red; (2) the sheep is drawn in green; and (3) the mirrored location that is being chased is drawn on the screen in blue -- whereas it was always invisible in the actual experiment. As a result, you can readily appreciate two things in the 'Cheat' version that are difficult to perceive in the first movie: the (red) wolf is clearly chasing the (blue) mirrored position, which is perfectly reflecting the motion of the (green) sheep through the center of the display.
The second task used in our paper -- the Don't Get Caught! task -- cannot be readily illustrated in movies in this fashion, since it is inherently interactive: in this task, the participant directly controls the sheep, and tries not to get caught. Here we illustrate two sample trials, though, to give the flavor of what the displays look like.
Experiment 5: Escaping from Wolves that Are or Are Not Facing You
Perfect heat-seeking (Chasing subtlety = 0 deg) (5.8 MB)
Here the wolf 'heat-seeks' the (moving) sheep -- with the wolf updating its position every 170 ms so as to be heading directly toward the sheep, which is being controlled by the subject. In this movie, the subject/sheep soon detects the wolf and then moves in a circular trajectory to avoid it. While watching, you may notice that you can soon detect the wolf, despite the large number of distractor objects. Note that this doesn't seem to be a result of intentional hypothesis-testing.
Large subtlety (Chasing subtlety = 90 deg) (2.4 MB)
Large subtlety (Chasing subtlety = 90 deg) [Cheat] (1.7 MB)
Here the direction in which the wolf moves is far less constrained, to within 90 deg of the subject's/sheep's true direction (for a total angular window of 180 deg, centered on the line connecting th wolf and sheep). Thus the wolf may now head (in the most extreme case) in an orthogonal direction to the (moving) sheep, but can never be heading away from it. In the first movie, the subject frantically tries to detect and avoid the wolf, but fails, and is caught; you may notice while watching this movie that the wolf is extremely difficult to detect. The second ('cheating') movie illustrates that the wolf is nevertheless effectively getting closer and closer to the sheep. Here the subject/sheep (now colored green) simply heads to a corner and stays put, and you can observe while the (red) wolf getting ever closer despite moving with the same 90 degrees of 'chasing subtlety'. (In this second movie, we've also explicitly used a larger blue circle around the subject/sheep to illustrate the 'kill zone' -- the region in which the wolf will automatically catch the sheep.)
This experiment adopted the Don't-Get-Caught task to explore the role of a different cue to the perception of animacy: the objects' directionality, operationalized as the way that the shape was 'facing' as it moved. (Chasing subtlety was fixed at 30 deg for all trials.) A large effect of this variable was observed, such that performance was much improved (in terms of subjects' ability to escape from the wolf) when the wolves and the other objects were always Oriented Darts that consistently pointed in the direction that they were moving. This effect is particularly striking in this experiment since this cue did not influence the actual efficacy of the wolves at chasing the user-controlled sheep (given that chasing subtlety was always equated across these conditions).
Heider & Simmel (1944) with Blocks (2.5 MB)
Heider & Simmel (1944) with Darts (2.5 MB)
These first demos, not used in our experiments, are analogues of the original movies of Heider and Simmel (1944), with the objects drawn as orientation-less shapes or as oriented darts. You may notice a richer and more immediate percept of animacy in the latter case.
Oriented Darts (5.2 MB)
Here the wolf and all distractors are drawn as darts that are always 'facing' the directions in which they are moving.
Misoriented Darts (5.1 MB)
Here the wolf and all distractors are drawn as darts whose orientations change haphazardly, unrelated to the directions in which they are moving. (The nature of their local rotations was perfectly controlled relative to those in the Oriented Darts condition, as described in the text.)
Perpendicular Darts (3.7 MB)
Here the wolf and all distractors are drawn as darts that are always 'facing' 90 degrees away from the directions in which they are moving.