Running Head: GOAL CHANGES CAUSED BY AGENTS AND PATIENTS
Action Goal Changes Caused by Agents and Patients Both Induce
the Updating of Event Models
Frank Papenmeier*, Annika Boss*, and Anne-Kathrin Mahlke
University of Tübingen, Germany
* These authors contributed equally to this work.
Author Note
Correspondence concerning this article should be addressed to Frank Papenmeier,
University of Tübingen, Schleichstr. 4, D-72076 Tübingen, Germany. E-mail:
[email protected]
Word count: 3753 (including references); Abstract: 237
GOAL CHANGES CAUSED BY AGENTS AND PATIENTS
Abstract
Observers represent everyday actions in event models along multiple dimensions such
as space, time or goals. Whenever new information along those dimensions is perceived, the
event model is updated accordingly. In the present research, we investigated goal changes
during ongoing actions, which involved both an agent performing the action and a patient
receiving the action. Previous research has found an agent advantage effect in visual
perception and event perception. In two experiments, we investigated whether goal changes
caused by an agent result in a more extensive updating of event models than goal changes
caused by a patient. We recorded short action clips showing goal-directed actions, such as
handing over a book. Those action sequences contained either no goal change or a goal
change caused by either the agent or the patient of the action. We generated image stills from
the action clips and participants viewed them as self-paced slideshows. This allowed us to
measure viewing times for each image still as dependent variable indicating processing time.
As predicted, images depicting the goal change caused an increased viewing time, thus
indicating an updating of event models across goal changes. However, the size of the
updating effect was comparable for goal changes caused by agents and goal changes caused
by patients. We conclude that observers update their event models when changes in goaldirected actions are perceived but that the updating is independent of the source of the goal
change.
Keywords: event model, agent advantage effect, viewing times, event updating, situation
model
GOAL CHANGES CAUSED BY AGENTS AND PATIENTS
Observers viewing everyday actions, such as viewing a customer handing over money
to a cashier, construct event models for the representation and comprehension of those
actions. Such event models are organized along a number of situational dimensions, such as
time, space, protagonist, causality and intentionality (Zwaan, Langston, & Graesser, 1995).
Whenever one of these dimensions changes, observers updated their event model (Zacks,
Speer, & Reynolds, 2009; Zwaan, Radvansky, Hilliard, & Curiel, 1998). Updating event
models is resource intensive and thus associated with increased processing times. This
applies not only to increased reading times during the comprehension of narratives (e.g.,
Curiel & Radvansky, 2014; Just & Carpenter, 1980; Zwaan, Magliano, & Graesser, 1995; but
see Radvansky & Copeland, 2010 for the limits of reading time experiments), but also to
increased viewing times during the comprehension of visual narratives and picture stories
(Cohn & Paczynski, 2013; Hard, Recchia, & Tversky, 2011; Magliano, Kopp, Higgs, &
Rapp, 2016; Magliano, Larson, Higgs, & Loschky, 2016). Thus, reading times and viewing
times often serve as a proxy for event model updating.
Whereas space and time changes received much attention by previous research (e.g.,
Curiel & Radvansky, 2014; Radvansky & Copeland, 2006, 2010; Rinck & Weber, 2003;
Zwaan, Magliano, et al., 1995), less research investigated goal changes. Goal changes are
associated with event model updating (Zwaan et al., 1998) and goal changes mediate the
perception of event boundaries (Speer, Zacks, & Reynolds, 2007). However, one important
aspect of goal changes had yet not been investigated: Changes in goal-directed actions can be
caused either by the agent performing the action or by the patient receiving the action.
Elaborating on the above example, the customer (agent) could cause a goal change by
retracting the money or the cashier (patient) could cause a goal change by refusing to take the
money. This distinction is theoretically highly relevant because a differential processing of
GOAL CHANGES CAUSED BY AGENTS AND PATIENTS
agent and patient information has been found for both visual perception and event perception
by previous research.
The so-called agent advantage effect describes a prioritized processing of agents in
visual perception. For example, observers are faster at identifying the agent than the patient
when presented with still images showing two fish with one biting the other (Segalowitz,
1982; Segalowitz & Hansson, 1979). Agents are also identified faster than patients in
dynamic events showing two shapes with one pushing the other (Verfaillie & Daems, 1996).
Importantly, recent research demonstrated that event models are constructed based on agent
information rather than patient information when viewing comic strips (Cohn & Paczynski,
2013). This conclusion was based on the findings that agents elicited more predictions about
the upcoming event and that actions were processed more quickly following the presentation
of the agent rather than the patient of the action.
In the present research, we studied the updating of event models following goal changes
caused by either the agent or the patient of an action. In particular, we were interested in
whether goal changes caused by the agent of an action result in stronger updating effects than
goal changes caused by the patient of an action. Given the prioritized processing of agent
information in visual perception and event perception, such an effect seemed plausible. In
contrast, however, updating the goal dimension of an event model could be indifferent to the
source of this change, which should then result in similar updating effects for both kinds of
goal changes.
Experiment 1
Method
Participants
GOAL CHANGES CAUSED BY AGENTS AND PATIENTS
Twenty-four students of the University of Tübingen (Mage = 23.42 years, SDage = 3.57)
participated in this experiment.
Apparatus and Stimuli
We recorded 28 action videos each showing two persons with one person (agent)
performing a goal-directed action towards another person (patient), such as hugging, kissing,
or handing over a book (see Table 1 for a complete list of all actions used). We recorded three
versions of each action: completed action, goal change caused by the agent (e.g., agent stops
to hand over a book and retracts it), or goal change caused by the patient (e.g., patient refuses
to grab the book that is handed over). This resulted in 84 action videos. For each action video,
we selected seven images best representing the action with the following restriction (see
Figure 1). The image in the action sequence that indicated that either no goal change or a goal
change was about to appear, henceforth called critical image, had to be at position four (6
actions), five (12 actions), or six (10 actions) of the action sequence. We introduced this
variability in order to prevent participants from learning that a specific position in the action
sequence might be special, thus preventing biased viewing times for that position. For each
action, the critical image was at the same positions for all conditions. Participants sat at an
unrestricted viewing distance of 65 cm to the display and images were presented with a size
of 22.2 x 12.6 degrees of visual angle.
Procedure and Design
We instructed the participants to watch the slide shows of the action sequences in a selfpaced manner (press space bar to continue to next image, inter-stimulus interval of 200 ms)
and to attend to their contents. After viewing the seven images of an action sequence,
participants rated the slide show for how easily comprehensible it was (1: hard to
comprehend – 7: easy to comprehend). We introduced this rating task in order to ensure that
GOAL CHANGES CAUSED BY AGENTS AND PATIENTS
participants comprehended the images and to distract the participants from the main measure
of the study, namely the viewing times for each slide. We presented all 84 sequences in a
randomized order.
Our manipulation resulted in a one-factorial within-subject design, with the independent
variable goal change (none, caused by agent, caused by patient) and the dependent variable
viewing time.
Results and Discussion
We removed the data of two action sequences („Kick“ and „Kick (version 2)“) from the
data set prior to analysis due to problems in the stimulus material (mis-ordered images and
visibility of goal change condition prior to the critical image). Furthermore, we considered
the first two trials of each participant as practice trials and excluded them from the analysis.
We applied the Greenhouse-Geisser correction to all results where the sphericity assumption
was violated as indicated by a significant Mauchly’s test.
We analyzed viewing times for the slides surrounding the critical image. Because the
critical image provided first information on either the completion or interruption of the
action, effects of goal change on viewing times should start to occur with the critical image.
We conducted a repeated-measures ANOVA with the factors time relative to critical image (2, -1, critical, +1) and goal change (none, caused by agent, caused by patient) and the
dependent variable viewing time (see Figure 2). As predicted, there was a significant
interaction of goal change and time relative to critical image, F(4.37, 100.62) = 6.96, p <
.001, ηp2 = .23. Further, there were significant main effects for goal change, F(2, 46) = 4.26, p
= .020, ηp2 = .16, and time relative to critical image, F(2.07, 47.68) = 32.18, p < .001, ηp2 =
.58. We further investigated the significant interaction with paired t-tests. Goal changes
caused by both agents and patients resulted in increased viewing times relative to the baseline
GOAL CHANGES CAUSED BY AGENTS AND PATIENTS
(no goal change) for the critical image and the image following the critical image, all ps ≤
.023, whereas viewing times did not differ significantly from baseline for the two images
preceding the critical image, all ps ≥ .065. Further, viewing times of the goal change caused
by agent condition and goal change caused by patient condition did not differ significantly for
all images, all ps ≥ .058. Thus, goal changes caused by agents and patients caused a similar
increase in viewing times indicating that comparable updating mechanisms occurred in both
conditions.
We analyzed the comprehensibility rating with a repeated measures ANOVA including
the factor goal change. This revealed a significant main effect of goal change, F(2, 46) =
147.51, p < .001, ηp2 = .87. Paired t-tests showed that all three goal change conditions
differed significantly from one another with the highest comprehensibility rating for
completed actions without goal change (M = 6.26, SD = 0.44), followed by goal changes
caused by patients (M = 4.64, SD = 0.84) and goal changes caused by agents (M = 3.50, SD =
0.89), all ps < .001. Whereas the comprehensibility rating was not the main measure of this
experiment, the reduced comprehensibility ratings for goal changes mimic our viewing time
results. In contrast to the viewing time measure, goal changes caused by agents were rated as
harder to comprehend than goal changes caused by patients.
Our main goal of this experiment was to study the viewing times for the first image
indicating the completion or interruption of an ongoing action. Therefore, we placed this
critical image at variable positions (4-6) across actions. As a side effect, some of the action
sequences contained only a single image following the critical image. Although the results of
this experiment indicate that the updating of event models is comparable across goal changes
caused by agents and patients, viewing times were still significantly increased in the image
following the critical image. Thus, it remains possible that updating processes associated with
goal changes caused by agents and patients diverge in later images. Therefore, we conducted
GOAL CHANGES CAUSED BY AGENTS AND PATIENTS
Experiment 2 in which we further investigated this question by including more images
following the critical image.
Experiment 2
With Experiment 2 we followed two aims. First, we included further images following
the critical image in order to study the development of viewing times following action goal
changes. In particular, we investigated whether differences in viewing times between goal
changes caused by agents and goal changes caused by patients might emerge in later images.
Second, we aimed at ensuring that the absence of differences in viewing times between agent
and patient goal changes are not the result of our stimulus material. Thus, we conducted a
pilot study prior to Experiment 2 and included only those action sequences in Experiment 2
that reliably elicited the correct event model (correct anticipated action prior to the critical
goal change image) in all three goal-change conditions.
Method
Participants
Twenty-four students of the University of Tübingen (Mage = 24.17 years, SDage = 4.85)
participated in this experiment. None of the participants had participated in Experiment 1.
Apparatus and Stimuli
We used the action sequences from Experiment 1 with the following changes. First, we
used 12 of the 28 action sequences (see Table 1). We selected those action sequences based
on a pilot study with 20 participants. In this pilot study, participants viewed all action
sequences as self-paced slide show, but action sequences stopped at the image preceding the
critical image. Participants’ task was to name the action that was going to be performed for
each sequence. We selected those action sequences where at least 90% of participants
correctly anticipated the performed action for the three goal change conditions. Thus, we
GOAL CHANGES CAUSED BY AGENTS AND PATIENTS
ensured to include only those action sequences where participants formed the same (and
intended) event model for all action conditions based on the images preceding the critical
image. As a second change, we added additional images to the action sequences in order to
ensure that every action sequence contained three image following the critical image. This
change allowed us to analyze the development of viewing times following the critical image.
Because the critical image was at different positions of the action sequences across actions,
action sequences no longer had a fixed length of seven images per sequence. Instead, the
action sequence consisted of seven pictures if the critical image was at position four (2
actions), it consisted of eight pictures if the critical image was at position five (4 actions), and
it consisted of nine pictures if the critical image was at position six (6 actions). The three
pictures following the critical image were chosen such that they best partitioned the
remaining action into three pieces.
Procedure and Design
The procedure and design of this experiment was the same as in Experiment 1.
Results and Discussion
As in Experiment 1, we considered the first two trials of each participant as practice
trials and excluded them from the analysis. Further, we applied the Greenhouse-Geisser
correction to all results where the sphericity assumption was violated as indicated by a
significant Mauchly’s test.
We conducted a repeated-measures ANOVA with the factors time relative to critical
image (-2, -1, critical, +1, +2, +3) and goal change (none, caused by agent, caused by patient)
and the dependent variable viewing time (see Figure 3). We replicated our findings from
Experiment 1. That is, there was a significant interaction of goal change and time relative to
critical image, F(4.67, 107.46) = 4.41, p = .001, ηp2 = .16. The main effect of time relative to
GOAL CHANGES CAUSED BY AGENTS AND PATIENTS
critical image was significant, F(2.73, 62.71) = 12.39, p < .001, ηp2 = .35, and the main effect
of goal change was not significant, F(1.57, 36.01) = 0.94, p = .381, ηp2 = .04. We further
investigated the significant interaction with paired t-tests. As in Experiment 1, goal changes
caused by both agents and patients resulted in increased viewing times relative to the baseline
(no goal change) for the critical image, both ps ≤ .017, whereas viewing times did not differ
significantly from baseline for the two images preceding the critical image, all ps ≥ .072.
Importantly, viewing times for goal changes did not differ significantly from baseline for the
two images following the critical image, all ps ≥ .141, and viewing times of the goal change
caused by agent condition and goal change caused by patient condition did not differ
significantly for all images, all ps ≥ .161. Thus, there was no evidence for the hypothesis that
the updating of event models takes longer for one of the two conditions. Instead, observers
completed the updating of their event model already within the critical image in this
experiment. This is different from Experiment 1 where we observed updating also in the
image following the critical image. Possibly, the inclusion of only a subset of the actions
from Experiment 1 explains this difference. Because Experiment 2 included only the action
sequences that reliably created a strong event model based on the images preceding the
critical image, event model updating could occur immediately whereas Experiment 1
included more variability. Further, it is worth to note that viewing times for goal changes
caused by agents and patients were longer than baseline in the final image of the action
sequence, both ps ≤ .009. Whereas this result is unexpected, it might be caused by the fixed
length of three images following the critical image in this Experiment. Presumably, the
critical image was particularly salient in the conditions with goal change and participants
prepared for their comprehensibility rating response in the last image of the action sequence.
We analyzed the comprehensibility rating with a repeated measures ANOVA including
the factor goal change. This revealed a significant main effect of goal change, F(2, 46) =
GOAL CHANGES CAUSED BY AGENTS AND PATIENTS
53.81, p < .001, ηp2 = .70. Paired t-tests showed that all three conditions differed significantly
from another with the highest comprehensibility rating for completed actions without goal
change (M = 6.27, SD = 0.50), followed by goal changes caused by patients (M = 5.08, SD =
1.02) and goal changes caused by agents (M = 4.04, SD = 1.29), all ps < .001. This replicates
the results of Experiment 1.
General Discussion
We conducted two experiments investigating the influence of goal changes introduced
by either the agent or the patient of an action on event model updating. We presented action
sequences as self-paced slideshows and measures viewing times. We observed a reliable
increase in viewing times following goal changes. However, the increase in viewing times
was comparable no matter whether the goal changes were caused by the agent or the patient
of an action.
The increase in viewing times following a goal change in our experiments mimics the
finding from research on narratives that found increased reading times following goal
changes (Zacks et al., 2009; Zwaan et al., 1998). Increased reading times are typically
interpreted as indicating event model updating (e.g., Curiel & Radvansky, 2014; Just &
Carpenter, 1980; Zwaan, Magliano, et al., 1995). This applies not only to reading but also to
the processing of picture stories (Cohn & Paczynski, 2013; Magliano, Kopp, et al., 2016;
Magliano, Larson, et al., 2016). Following this tradition, we conclude that goal changes
caused an updating of event models in our experiments and that the updating effect was
comparable for goal changes caused by agents and patients.
Previous research found an agent advantage effect, namely a faster visual processing of
agents than patients (Segalowitz, 1982; Segalowitz & Hansson, 1979) and the construction of
event models based on agents rather than patients (Cohn & Paczynski, 2013). Based on these
GOAL CHANGES CAUSED BY AGENTS AND PATIENTS
findings, we hypothesized that goal changes caused by agents might cause a stronger
updating effect than goal changes caused by patients. However, both goal changes caused
comparable updating in our experiments. Therefore, our findings are consistent with the idea
that event models are indexed along a number of dimensions (Zwaan, Langston, et al., 1995)
and that changes in those dimensions (such as goals or intentions) cause an updating of the
respective dimension in the event model. For this to happen, it is irrelevant how the
dimension change occurred. Following this argument, future research should investigate
multi-step actions where an action interruption by a patient (e.g., avoiding a punch) is
followed by the patient turning in the agent (e.g., punching the former agent). Such a change
in the protagonist dimension should then cause further increases in viewing times and event
model updating as compared with conditions where the agent is again acting on the patient.
In our experiments, we created picture sequences of short action videos showing a
single ongoing action in order to measure event model updating in the midst of an ongoing
event. Doing so, we were able to measure increased processing times at the very moment of
the action goal change. Previous research also demonstrated increased viewing times for
images perceived as an event boundary (Hard et al., 2011). Thus, we propose that future
research should apply this method to the study of event model updating across changes in
situational dimensions (e.g., time, space, protagonist, causality and intentionality; Zwaan,
Langston, et al., 1995) in visual narratives and movies. Because viewing times provide a
continuous measure of processing effort and thus the amount of event model updating, such
an approach might be able to resolve the debate on whether event model updating is
incremental, global, or a mixture of both (Bailey & Zacks, 2015; Gernsbacher, 1997; Huff,
Meitz, & Papenmeier, 2014; Kurby & Zacks, 2012).
Taken together, we found that goal changes during a single ongoing action caused event
model updating as shown by increased viewing times on respective images. Surprisingly
GOAL CHANGES CAUSED BY AGENTS AND PATIENTS
however, it did not make any difference whether the goal changes were caused by either the
agent or patient of an action. Thus, we conclude that changes in the goal dimensions cause
event model updating irrespective the source of the goal change.
Author Contributions
All authors developed the study concept and contributed to the study design. A. Boss
recorded the action sequences and A. Mahlke and F. Papenmeier programmed the experiment.
All authors were responsible for testing, data collection, and performed the data analysis. F.
Papenmeier and A. Boss drafted the manuscript, and A. Mahlke provided critical revisions.
All authors approved the final version of the manuscript for submission.
Author Note
We thank Annika Thierfelder and Svenja Leonie Brosch for their help in conducting
Experiment 1 and Eleni Sianni, Lisa Krösche, and Johanna-Lowis Donath for their help in
conducting Experiment 2. We provide all data collected for the experiments presented in this
manuscript as open data at the following location:
https://osf.io/wtkuy/?view_only=65841b760f3f4093859fa9dc0ba2351f
GOAL CHANGES CAUSED BY AGENTS AND PATIENTS
References
Baguley, T. (2012). Calculating and graphing within-subject confidence intervals for
ANOVA. Behavior Research Methods, 44, 158–175. https://doi.org/10.3758/s13428011-0123-7
Bailey, H. R., & Zacks, J. M. (2015). Situation model updating in young and older adults:
Global versus incremental mechanisms. Psychology and Aging, 30, 232–244.
https://doi.org/10.1037/a0039081
Cohn, N., & Paczynski, M. (2013). Prediction, events, and the advantage of Agents: The
processing of semantic roles in visual narrative. Cognitive Psychology, 67(3), 73–97.
https://doi.org/10.1016/j.cogpsych.2013.07.002
Curiel, J. M., & Radvansky, G. A. (2014). Spatial and character situation model updating.
Journal of Cognitive Psychology, 26(2), 205–212.
https://doi.org/10.1080/20445911.2013.879590
Gernsbacher, M. A. (1997). Two decades of structure building. Discourse Processes, 23, 265–
304. https://doi.org/10.1080/01638539709544994
Hard, B. M., Recchia, G., & Tversky, B. (2011). The shape of action. Journal of
Experimental Psychology: General, 140, 586–604. https://doi.org/10.1037/a0024310
Huff, M., Meitz, T. G. K., & Papenmeier, F. (2014). Changes in situation models modulate
processes of event perception in audiovisual narratives. Journal of Experimental
Psychology: Learning, Memory, and Cognition, 40, 1377–1388.
https://doi.org/10.1037/a0036780
Just, M. A., & Carpenter, P. A. (1980). A theory of reading: From eye fixations to
comprehension. Psychological Review, 87(4), 329–354. https://doi.org/10.1037/0033295X.87.4.329
GOAL CHANGES CAUSED BY AGENTS AND PATIENTS
Kurby, C. A., & Zacks, J. M. (2012). Starting from scratch and building brick by brick in
comprehension. Memory & Cognition, 40, 812–826. https://doi.org/10.3758/s13421011-0179-8
Magliano, J. P., Kopp, K., Higgs, K., & Rapp, D. N. (2016). Filling in the gaps: Memory
implications for inferring missing content in graphic narratives. Discourse Processes,
1–14. https://doi.org/10.1080/0163853X.2015.1136870
Magliano, J. P., Larson, A. M., Higgs, K., & Loschky, L. C. (2016). The relative roles of
visuospatial and linguistic working memory systems in generating inferences during
visual narrative comprehension. Memory & Cognition, 44, 207–219.
https://doi.org/10.3758/s13421-015-0558-7
Radvansky, G. A., & Copeland, D. E. (2006). Walking through doorways causes forgetting:
Situation models and experienced space. Memory & Cognition, 34(5), 1150–1156.
Radvansky, G. A., & Copeland, D. E. (2010). Reading times and the detection of event shift
processing. Journal of Experimental Psychology: Learning, Memory, and Cognition,
36, 210–216. https://doi.org/10.1037/a0017258
Rinck, M., & Weber, U. (2003). Who when where: An experimental test of the eventindexing model. Memory & Cognition, 31(8), 1284–1292.
https://doi.org/10.3758/BF03195811
Segalowitz, N. S. (1982). The perception of semantic relations in pictures. Memory &
Cognition, 10(4), 381–388.
Segalowitz, N. S., & Hansson, P. (1979). Hemispheric functions in the processing of agentpatient information. Brain and Language, 8, 51–61. https://doi.org/10.1016/0093934X(79)90039-7
GOAL CHANGES CAUSED BY AGENTS AND PATIENTS
Speer, N. K., Zacks, J. M., & Reynolds, J. R. (2007). Human brain activity time-locked to
narrative event boundaries. Psychological Science, 18(5), 449–455.
https://doi.org/10.1111/j.1467-9280.2007.01920.x
Verfaillie, K., & Daems, A. (1996). The priority of the agent in visual event perception: On
the cognitive basis of grammatical agent-patient asymmetries. Cognitive Linguistics,
7, 131–147.
Zacks, J. M., Speer, N. K., & Reynolds, J. R. (2009). Segmentation in reading and film
comprehension. Journal of Experimental Psychology: General, 138, 307–327.
https://doi.org/10.1037/a0015305
Zwaan, R. A., Langston, M. C., & Graesser, A. C. (1995). The construction of situation
models in narrative comprehension: An event-indexing model. Psychological Science,
6(5), 292–297. https://doi.org/10.1111/j.1467-9280.1995.tb00513.x
Zwaan, R. A., Magliano, J. P., & Graesser, A. C. (1995). Dimensions of situation model
construction in narrative comprehension. Journal of Experimental Psychology:
Learning, Memory, and Cognition, 21, 386–397. https://doi.org/10.1037/02787393.21.2.386
Zwaan, R. A., Radvansky, G. A., Hilliard, A. E., & Curiel, J. M. (1998). Constructing
multidimensional situation models during reading. Scientific Studies of Reading, 2(3),
199–220. https://doi.org/10.1207/s1532799xssr0203_2
GOAL CHANGES CAUSED BY AGENTS AND PATIENTS
Figure Captions
Figure 1. Illustration of a typical action sequence shown in Experiment 1. Each action
sequence consisted of seven pictures and there were three versions of each action sequence:
No goal change, goal change caused by agent, and goal change caused by patient. In this
example, this corresponds to handing over a glass, agent starts to hand over a glass but
retracts it, and agent starts to hand over a glass but the patient refuses to take it. The critical
image was the first image indicating either the completion or interruption of the action. In this
example, the fourth image was defined as the critical image.
Figure 2. Results of Experiment 1. Goal changes caused by both agents and patients resulted
in a comparable increase in viewing times beginning with the critical image showing the goal
change. Error bars indicate 95% within-subject confidence intervals (Baguley, 2012).
Figure 3. Results of Experiment 2. Goal changes caused by both agents and patients resulted
in a comparable increase in viewing times at the critical image showing the goal change.
Error bars indicate 95% within-subject confidence intervals (Baguley, 2012).
GOAL CHANGES CAUSED BY AGENTS AND PATIENTS
Table Captions
Table 1
List of actions used in our experiments. For some actions, we recorded two versions with
different actors.
GOAL CHANGES CAUSED BY AGENTS AND PATIENTS
Figure 1. Illustration of a typical action sequence shown in Experiment 1. Each action
sequence consisted of seven pictures and there were three versions of each action sequence:
No goal change, goal change caused by agent, and goal change caused by patient. In this
example, this corresponds to handing over a glass, agent starts to hand over a glass but
retracts it, and agent starts to hand over a glass but the patient refuses to take it. The critical
image was the first image indicating either the completion or interruption of the action. In this
example, the fourth image was defined as the critical image.
GOAL CHANGES CAUSED BY AGENTS AND PATIENTS
Figure 2. Results of Experiment 1. Goal changes caused by both agents and patients resulted
in a comparable increase in viewing times beginning with the critical image showing the goal
change. Error bars indicate 95% within-subject confidence intervals (Baguley, 2012).
GOAL CHANGES CAUSED BY AGENTS AND PATIENTS
Figure 3. Results of Experiment 2. Goal changes caused by both agents and patients resulted
in a comparable increase in viewing times at the critical image showing the goal change.
Error bars indicate 95% within-subject confidence intervals (Baguley, 2012).
GOAL CHANGES CAUSED BY AGENTS AND PATIENTS
Table 1
List of actions used in our experiments. For some actions, we recorded two versions with
different actors.
Experiments 1 and 2
Experiment 1 only
Actions
Draw Blood; Handshake; Hand Over Glass; Help Somebody Up; High Five; High
Five (version 2); Hugging; Hugging (version 2); Injection; Mouth Swab; Put Arm
Around Others' Shoulder; Slap Hands
Hand Over Beer; Hand Over Book; Hand Over Book (version 2); Kick; Kick
(version 2); Kiss; Kiss On Others’ Forehead; Slap in the Face; Slap in the Face
(version 2); Pat On Shoulder; Pat On Shoulder (version 2); Punch; Punch (version
2); Push; Push (version 2); Put Arm Around Others' Shoulder (version 2)