2022 - Effects of Spatiotemporal (Dis) Continuity On Working Memory For Human Movements

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Acta Psychologica 230 (2022) 103719

Contents lists available at ScienceDirect

Acta Psychologica
journal homepage: www.elsevier.com/locate/actpsy

Effects of spatiotemporal (dis)continuity on working memory for


human movements
Shiau-Chuen Chiou *
Neurocognition and Action Research Group, Center for Cognitive Interaction Technology (CITEC), Bielefeld University, Bielefeld, Germany
Faculty of Psychology and Sports Science, Bielefeld University, Bielefeld, Germany

A R T I C L E I N F O A B S T R A C T

Keywords: Human movements are dynamic and continuous in nature. However, how the spatiotemporal continuity in­
Visual perception fluences working memory for movements is still unclear. Specifically, spatiotemporal continuity of movements
Working memory may facilitate integrative processing (“integration”) and enhance memory performance by optimizing the
Spatial
encoding process, but it may also diminish memory benefits from distinctive processing (“separation”). In this
Temporal
Continuity
study, we manipulated the continuity state (continuous/discontinuous) (Experiment 1) and its predictability
Whole-body movement (Experiment 2) of whole-body movement sequences and tested participants' working memory for observed
movements with a single-probe recognition task. We formulated potential influence from spatiotemporal (dis)
continuity by two opposite forces — integration vs. separation, and demonstrated a conflict between these two
processes across space and time. Moreover, we found that the seemingly stimulus-driven perceptual effects from
spatiotemporal (dis)continuity might be supported by a prediction-based mechanism, which guided the selection
of an optimal processing strategy. Overall, our finding illustrates an interweaving relationship between spatial
and temporal processing during action observation and highlights the importance of considering the dynamic
and continuous nature of human movements in visual perception and working memory research.

1. Introduction example, Gunnar Johansson (1973, 1976) suggested that the recogni­
tion of biological motions is guided by relative motions of individual
From learning a short series of steps in a dance studio to under­ body parts, which are perceptually organized to a gestalt. More recent
standing a sequence of motor actions performed by other individuals, it research also demonstrated higher-level global processing of biological
is an essential ability of humans to store a sequence of movements in a motions in both spatial (Bertenthal & Pinto, 1994; Shiffrar et al., 1997)
temporary buffer, known as working memory, to facilitate subsequent and temporal (Thornton et al., 1998) domains, illustrating potential
behavioral responses. Different from static objects that normally occupy benefits from integrative processing on movement perception.
isolated positions in space, human movements unfold over time and are Moreover, spatiotemporal continuity may also increase the predict­
continuous in nature. This “spatiotemporal continuity” is not only an ability of movements. For one thing, continuous movements better
intrinsic characteristic of human movements, but also serves as a type of follow the anatomical constraints and biomechanical dynamics of
contextual information that may influence how movement information human motor system than discontinuous ones; for another thing, the
is encoded and retained in working memory. spatiotemporal dependency between movement units further restricts a
Specifically, spatiotemporal continuity may enhance perceptual theoretically infinite number of ways of movement combinations to
integration of movements through Gestalt principles of grouping, such limited ones. According to the event segmentation theory (Zacks et al.,
as proximity (Wertheimer, 1923), connectedness (Palmer & Rock, 2007), people tend to divide a continuous perceptual experience into
1994), or continuity (see Wagemans et al., 2012 for a review). In object parts, i.e., segmentation, also known as the idea of unitization —
recognition, elements that share a common border tend to be recognized grouping some of a continuum into a gestalt (Zacks, 2020). Event seg­
as being part of the same object. Similarly, movements that happen close mentation is mainly supported by predictive processing, with an event
in time, connected in space, or with good continuity in configuration are boundary being perceived when a transient increase of prediction error
likely to be perceived as belonging to the same movement or event. For occurs (i.e., a failure in predicting what will happen next) (Richmond &

* Center for Cognitive Interaction Technology (CITEC), Bielefeld University, Inspiration 1, 33619 Bielefeld, Germany.
E-mail address: shiau-chuen.chiou@uni-bielefeld.de.

https://doi.org/10.1016/j.actpsy.2022.103719
Received 30 April 2022; Received in revised form 20 July 2022; Accepted 18 August 2022
Available online 23 August 2022
0001-6918/© 2022 The Author. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
S.-C. Chiou Acta Psychologica 230 (2022) 103719

Zacks, 2017; Zacks et al., 2007). As being more predictable, a continuous probe recognition task (Sternberg, 1969) (Experiment 1). The item-
movement sequence is also more likely to be perceived as a single event based retrieval required by the task was assumed to force participants
or an integrated “whole” than a discontinuous sequence. Taken to process individual item information even when global-level repre­
together, the spatiotemporal continuity of movements may facilitate the sentations were formed. In addition, to investigate the extent to which
integrative processing of incoming information through both stimulus- the prediction-based mechanism (as opposed to the stimulus-driven
driven Gestalt principles and knowledge-based predictions. mechanism) might contribute to the effects of spatiotemporal (dis)con­
But how does the integrative processing influence working memory? tinuity observed in Experiment 1, we abolished the predictability of the
Previous research suggests that combining multiple elements into a continuity state and examined how this might influence the presence or
higher-order description such as a perceptual group or a chunk (Miller, absence of the (dis)continuity effects (Experiment 2).
1956; Nassar et al., 2018) can enhance memory performance, as it op­ As discussed previously, integration effects, if any, should be stron­
timizes the encoding process, enables more efficient memory represen­ ger when observing continuous than discontinuous sequences, while
tations, and thus reduces the overall load on memory (Brady et al., 2009, separation (or isolation) effects, if present, should be larger when
2011; Cowan, 2001; Cowan et al., 2004; Xu & Chun, 2007). For example, observing discontinuous than continuous sequences. Therefore, a posi­
it is easier to memorize positions of multiple dots if they form a familiar tive effect when contrasting performance in continuous with that in
spatial configuration. However, integrative processing (“integration”) discontinuous conditions would indicate a stronger influence from
may diminish memory benefits from distinctive processing (“separa­ integration over separation; a negative effect would suggest a larger
tion”). Here, “integration” is defined as a process underlying any contribution from separation over integration, and a null effect would
possible global-level representations enabled or facilitated by the imply either that there is no integration or separation effect when
spatiotemporal continuity of movements, while “separation” is defined observing continuous or discontinuous sequences or that these two ef­
as the opposite process of integration and can be understood, in this fects contribute equally to the process and thus offset against each other.
context, as a concept interchangeable with “distinctiveness” (i.e.,
distinctive processing, rather than an inherent property of to-be- 2. Experiment 1
remembered materials) (Hunt & Worthen, 2006), “isolation” (e.g.,
Morin et al., 2010), or “segmentation” (Zacks et al., 2007). 2.1. Method
Separation has been shown to facilitate memory because it precisely
specifies items within events and reduces interference during retrieval 2.1.1. Participants
(see Hunt & Worthen, 2006 for a review; Morin et al., 2010; Pettijohn We recruited 24 participants for the experiment (12 female; aged
et al., 2016). For example, temporal distinctiveness models of working 18–40 years, M = 24.3, SD = 5.0). No participants have viewed the
memory (Brown et al., 2000; Brown et al., 2007; Glenberg & Swanson, experimental materials before. Data of three participants were replaced
1986; Neath & Crowder, 1996) suggest that temporally isolated items by new participants due to a performance out of 2 standard deviations
are recalled better, as their memory traces are more discriminable by from the group mean and a poor response quality inspected by the
occupying relatively isolated positions along the temporal dimension in experimenter (e.g., making responses before stimulus onset, making
memory. The so-called “temporal isolation effect” is typically found in more than one response in the same trial, etc.). A sample size of 24 was
recognition (Morin et al., 2010) and free recall tasks (Brown et al., 2006; determined based on a power analysis (using G*Power 3.1; Faul et al.,
Glenberg & Swanson, 1986; Neath & Crowder, 1996), but not in forward 2007) to provide a power of 0.80 at an alpha level of 0.05 to detect
serial recall (Lewandowsky et al., 2006; Nimmo & Lewandowsky, 2005; moderate- to large-sized effects of integration and/or separation (d =
Parmentier et al., 2006), in which the report order of items is strictly 0.6) (Oberauer et al., 2018) for paired-samples, within-subjects
forward, reducing potential influence from time (Geiger & Lew­ comparisons.
andowsky, 2008; Lewandowsky et al., 2008). Participants' experiences in dance, music, and sport were evaluated
Note that the temporal distinctiveness account is not the only hy­ by a questionnaire and reported here as “expertise indexes” (0: No
pothesis to explain the beneficial effect of time for working memory. experience, 1: Beginner, 2: Intermediate amateur, 3: Advanced amateur,
Presenting items with temporal gaps may also provide a chance to 4: Professional) of 0.3 (SD = 0.5), 0.6 (SD = 0.8), and 1.4 (SD = 1.0),
consolidate the encoded information, leading to a slower rate of respectively, defined by both the training length and the self-evaluated
forgetting. For example, Ricker and Cowan (2014) showed that skill level.1 No professionals were recruited in the present research.
sequential presentation of memory items, which provided extra time for Participants signed informed consent prior to the experiment and
consolidation after the presentation of each item, led to better perfor­ received €8 per hour for their participation. Experiments presented in
mance than simultaneous presentation. Lewandowsky and Brown this paper were conducted in accordance with the ethical principles
(2005) also demonstrated that the duration of post-item intervals (but stated within the declaration of Helsinki (1964) and were approved by
not pre-item intervals) played a role in recall accuracy, indicating ben­ the Ethics Committee of Bielefeld University.
efits from memory consolidation rather than pure isolation. Moreover,
temporal gaps between item presentation may also allow more encoding 2.1.2. Stimuli and apparatus
resource to be replenished (Mizrak & Oberauer, 2021; Popov & Reder, Thirty whole-body movement sequences were used in the experi­
2020), facilitating the encoding of succeeding units. In this paper, we ment (half performed by a female dancer and the other half performed
used the term “temporal isolation effect” to refer to memory benefits by a male dancer, both in fitted black clothing). Each sequence was
from temporally isolated items, i.e., items presented with temporal gaps, composed of four movement units, all without interpretable external
with no prior assumptions about its underlying mechanism. goals and action semantics. We defined spatial continuity of a movement
Overall, although the spatiotemporal continuity of movements may sequence in its loosest form by linking the ending pose of the first unit
facilitate integration and enhance memory performance by optimizing with the starting pose of the second one, and so on, while preserving the
the encoding process, it may also diminish memory benefits from sep­
aration. In this study, we formulated potential influence from spatio­
temporal (dis)continuity by these two opposite forces — integration vs. 1
The expertise index (0–4) is defined as the following: 0: No experience, 1:
separation, and examined how they interactively act upon memory
Beginner (skill level = 1 or 2 in a five-point scale, or skill level > 2 but training
process and influence memory performance. Specifically, we manipu­ length < 3–5 years), 2: Intermediate amateur (skill level = 3 and training
lated the continuity state (continuous/discontinuous) of whole-body length ≥ 3–5 years, or skill level > 3 but training length < 6 years), 3: Advanced
movement sequences in both spatial and temporal domains and tested amateur (skill level = 4 and training length ≥ 6 years), 4: Professionals (skill
participants' working memory for observed movements with a single- level = 5 and training length ≥ 6 years).

2
S.-C. Chiou Acta Psychologica 230 (2022) 103719

discrete (self-sufficient) nature of individual units. Specifically, each sequence; the other half were negative trials on which the probe was
unit of a sequence had its own bell-shaped velocity profile (i.e., accel­ absent from the sample sequence but chosen randomly from the stimulus
erating till the midpoint of the movement and then decelerating) (Abend set without replacement (see Videos S2 and S3 in Supplementary in­
et al., 1982), yielding a clear starting point and ending point where the formation for sample video clips that illustrate positive and negative
velocity was zero. Moreover, each movement unit had a unique pattern probes, respectively, when Video S1 is the sample sequence).
and the use of body parts, and would form a continuous trajectory with Each trial began with a 1-s fixation cross (+), followed by a sample
the adjacent units only when aligned in a specific order (see Fig. 1 for an sequence and a mask with a black-white chessboard pattern for 1.5 s.
illustration and Video S1 in Supplementary information for a sample Offset of the mask was a neutral response cue (a gray circle, 0.6 degree in
video clip). diameter) for 700 ms, followed by a single movement unit as the
Each sequence was eight beats long (two beats per movement unit) in recognition probe and a question “Is this movement part of the sequence?”
a 4/4 musical meter and paced at a tempo of 90 beats per minute, Participants were required to make a yes/no judgment by keystroke on a
yielding a sequence length of about 6 s after including one additional standard computer keyboard (the position of “F” key and “J” key,
beat for preparation at the beginning of the sequence. Movement re­ respectively, marked in red) as fast and accurate as possible, but no strict
cordings were made with a digital video camera recorder (Sony HDR- response time limit was imposed (Fig. 2). In practice trials, participants
CX430V) at 50 frames per second against a white background and a were allowed to replay the videos before making a judgment and
gray floor. Videos were edited on a frame basis using the software received feedback (as shown by the word correct or incorrect on the
iMovie (Apple, Inc.) and were displayed silently to participants at 1600 computer screen) on a trial-by-trial basis. In addition, to ensure partic­
× 900 pixels on a 24-in. LCD screen (Dell U2412M) with a viewing ipants' understanding of the task, incorrect trials would be replayed once
distance of approximately 50 cm. The experimental flow and data automatically with an oral check by the experimenter if necessary. No
analysis were programmed in Python; stimuli presentation was imple­ video replay or feedback was provided in formal experiment.
mented with the PsychoPy software package (Peirce, 2007, 2009).
2.1.4. Data analysis
2.1.3. Design and procedure Overall performance of the recognition task was measured by hit rate
We used a within-subjects two-by-two factorial design, of which (correctly responding “yes” on positive trials) minus false alarm rate
Spatial Continuity (continuous/discontinuous) and Temporal Continu­ (incorrectly responding “yes” on negative trials), denoted as Hits – FAs
ity (continuous/discontinuous) of movement sequences were the two (Rouder et al., 2011). If Hits – FAs < 0, performance was set to be 0. In
independent variables. Participants performed a single-probe recogni­ addition, to better understand how an observed movement sequence
tion task under these four trial conditions in separate sessions: (1) might be represented in memory and how the input serial position might
spatial-continuous + temporal-continuous, (2) spatial-continuous + influence memory performance, accuracy (hit rate) and response times
temporal-discontinuous, (3) spatial-discontinuous + temporal- (RTs) for positive trials were analyzed as a function of input serial po­
continuous, and (4) spatial-discontinuous + temporal-discontinuous. sition. Here we included positive trials only as negative probes did not
Twenty-four possible orders of performing these four conditions were have input serial position. RTs were recorded from the onset of the
randomly assigned to the twenty-four participants of the experiment. recognition probe until the next key press, and RTs associated with er­
Movement sequences were originally recorded as spatiotemporal- rors or exceeding two standard deviations from the individual means
continuous sequences (see Fig. 1). Spatial-discontinuous sequences were eliminated.
were then created by switching the display order of the second and the Statistical threshold of Type I error was set at α = 0.05; Cohen's d and
third units of spatial-continuous sequences, and temporal-discontinuous partial eta squared (η2p) were reported to indicate effect size. Post hoc
sequences were created by inserting inter-stimulus intervals (ISIs) of 0.5 analyses were conducted using Bonferroni correction, and Greenhouse-
s, 1.0 s, or 1.5 s, shown as black screens, between movement units (see Geisser correction was applied when the assumption of sphericity had
Fig. 2). been violated in an analysis of variance (ANOVA).
At the beginning of the experiment, a general introduction was
provided both literally and verbally with graphic illustrations. In each
session, participants performed 6 practice trials (with movement se­ 2.2. Results
quences not used in formal experiment) and 60 experimental trials (20
trials/block) (see below for trial design). Half of the trials were positive 2.2.1. Overall performance: Hits – FAs
trials on which the probe movement has been displayed in the sample We first conducted a 2 (Spatial Continuity: continuous, discontin­
uous) × 2 (Temporal Continuity: continuous, discontinuous) repeated-

Fig. 1. Illustration of a whole-body movement sequence. Movement sequences started at the center of the recording region with a relaxed standing pose, followed by
four linked movement units, i.e., the ending pose of the first unit was the starting pose of the second one, and so on. Movements were all without interpretable
external goals and action semantics (see Video S1 in Supplementary information for a sample video clip). Reprinted from Chiou (2022) (CC BY 4.0).

3
S.-C. Chiou Acta Psychologica 230 (2022) 103719

Fig. 2. Trial structure of the recognition task and sequence design of the four trial conditions in Experiment 1. Spatial-discontinuous sequences (1324) were created
by switching the display order of the second and the third units of spatial-continuous sequences (1234). Temporal-discontinuous sequences were created by inserting
inter-stimulus intervals (ISIs) of 0.5 s, 1.0 s, or 1.5 s, shown as black screens, between movement units. ITI indicates inter-trial interval.

measures ANOVA on Hits – FAs. The results yielded no significant effect Moreover, the results also showed that movement units of a spatial-
for Spatial Continuity, F(1, 23) = 3.04, p = .095, η2p = 0.12, Temporal continuous sequence (M = 75.1 %, SD = 15.4 %) were memorized better
Continuity, F(1, 23) = 1.77, p = .196, η2p = 0.07, or the two-way inter­ in average than those of a spatial-discontinuous sequence (M = 69.7 %,
action, F(1, 23) = 2.21, p = .150, η2p = 0.09, seeming to suggest that the SD = 17.3 %) when ISIs were shorter than 1.5 s (i.e., when spatial-
spatiotemporal (dis)continuity had no influence on working memory discontinuous sequences had yet to benefit from temporal isolation), t
performance. However, there was at least one more factor that deserved (23) = 2.68, p = .013, d = 0.55, illustrating memory benefits from
further scrutiny, namely the ISIs applied to the temporal-discontinuous integrative processing in the spatial domain (Fig. 3). To provide a
condition. Specifically, ISIs of 1 s in average might be too short to induce quantitative measure of working memory capacity, we adopted Cowan's
observable isolation effects. (2001) formula to estimate the number of movement units that can be
To check this possibility, we examined the fine-grained effects from retained in working memory: k = n × (Hits – FAs), in which k is the
temporal (dis)continuity by conducting a 2 (Spatial Continuity: contin­ estimated working memory capacity and n is the number of movement
uous, discontinuous) × 4 (ISI: 0 s, 0.5 s, 1.0 s, 1.5 s) repeated-measures units in a sequence (i.e., set size) (see also Rouder et al., 2011). Based on
ANOVA. The results showed no main effect for Spatial Continuity, F(1, this approach, the data showed that participants were able to retain 3.01
23) = 1.48, p = .236, η2p = 0.06, or ISI, F(3, 69) = 1.64, p = .189, η2p = movement units of a spatial-continuous sequence and 2.79 movement
0.07, but a significant interaction effect between the two, F(3, 69) = units of a spatial-discontinuous sequence when ISIs were shorter than
4.44, p = .007, η2p = 0.16 (Fig. 3). Post hoc analyses yielded that per­ 1.5 s. The finding is consistent with previous research showing that the
formance improved significantly in the spatial-discontinuous condition working memory capacity of movements is about two to four units (Shen
when ISIs increased, F(2.20, 50.6) = 5.56, p = .005, η2p = 0.20. Move­ et al., 2014; Wood, 2007).
ment units were memorized better when displayed with temporal gaps
of 1.5 s (M = 80.0 %, 95 % CI [73.0 %, 87.0 %]) as compared with 0 s (M 2.2.2. Serial-position effects: accuracy (hit rate) and RTs for positive trials
= 69.2 %, 95 % CI [62.5 %, 75.9 %]), t(23) = 3.93, p = .004, d = 0.80, or To investigate how an observed movement sequence might be rep­
0.5 s (M = 71.7 %, 95 % CI [63.7 %, 79.6 %]), t(23) = 3.50, p = .012, d resented in memory and how the input serial position might influence
= 0.71. However, performance in the spatial-continuous condition memory performance, we analyzed accuracy (hit rate) and RTs for
remained constant across all ISIs, F(3, 69) = 0.14, p = .939, η2p = 0.006. positive trials as a function of input serial position. A one-way repeated-
The results indicated that the temporal isolation effects were only pre­ measures ANOVA on hit rate revealed a significant main effect for serial
sent in the spatial-discontinuous condition when ISIs were long enough position, F(1.84, 42.4) = 15.1, p < .001, η2p = 0.40. Post hoc analyses
(1.5 s), but not in the spatial-continuous condition. indicated a recency effect but no primacy effect. The last unit of a
sequence was recognized better than all previous units, ps ≤ 0.001,
while performance for the first three units did not show significant
difference, ps ≥ 0.486 (Fig. 4a). The analysis on RTs yielded similar
results. There was a significant main effect for serial position, F(3, 69) =
42.8, p < .001, η2p = 0.65. RTs were the shortest when the probed
movement was the last unit of a sequence, ps < 0.001, indicating a
recency effect, while RTs for the first unit was longer than those for the
second, p < .001, or the third, p = .007, unit, suggesting no primacy
effect. (Fig. 4b).

2.3. Discussion

In Experiment 1, we manipulated the continuity state of whole-body


movement sequences and investigated how the spatiotemporal (dis)
continuity influences working memory for movements. Overall, we
found (i) a stronger effect from integration over separation in the spatial
domain (“spatial integration”) given temporal continuity and (ii) a
Fig. 3. Memory performance (measured by Hits – FAs) in Experiment 1 as a larger influence from separation over integration in the temporal
function of inter-stimulus interval. Error bars indicate 95 % confidence in­ domain (“temporal isolation”) given spatial discontinuity.
tervals calculated across participants. Note that temporal isolation effects did not occur in the spatial-

4
S.-C. Chiou Acta Psychologica 230 (2022) 103719

Fig. 4. Serial-position effects in Experiment 1 (after collapsing ISIs) when measured by hit rate (a) and response times (RTs) (b). Error bars indicate 95 % confidence
intervals calculated across participants.

continuous condition. One possibility is that the spatial dependency differentiating between movement units with the same starting posture
between movement units (or more precisely the integration of adjacent (i.e., movement unit 1, as it always started with a relaxed standing
movement units) reduced effective isolation in the temporal domain. preparation posture) might be more difficult than differentiating be­
Specifically, participants might prolong mental representations of indi­ tween those with different starting postures. However, the same starting
vidual movement units to “fill the gaps” if it would lead to a continuous posture of movement unit 1 might also simplify the decision process, as
percept (integration process). Previous studies on biological motion the recognition probe, if starting with the preparation posture, only
perception have shown that participants actively bridged the temporal needed to be compared with the first unit of the sample sequence. Since
gaps between two successive body postures to create a continuous the unit information was collapsed in determining the effect of spatio­
percept of movement (Chatterjee et al., 1996; Orgs et al., 2011; Shiffrar temporal (dis)continuity, the inconsistency of the starting posture would
& Freyd, 1990, 1993). Evidence of representational momentum (Freyd not influence the interpretation of the main results. Concerning the
& Finke, 1984; Hubbard, 2005), which showed the tendency of an serial-position effect, although the current design was not sufficient to
observer to remember the final position of a motion (or an implied determine its influence on the recognition performance of movement
motion) as extending beyond its actual position, also supports this unit 1, a clear upward trend on accuracy and downward trend on RTs
possibility. However, if two successive body postures were apparently observed in serial position curves (see Fig. 4 and also Fig. 7 in Experi­
from different movements (i.e., spatially discontinuous), participants ment 2) would still suggest the one-item recency effect (with limited
would have no incentive to form sustained representations of move­ primacy effect, if any).
ments, but would instead make use of the gaps to form effective isolation In sum, the finding of Experiment 1 illustrates a trade-off between
(separation process). spatial integration and temporal isolation and demonstrates a potential
Alternatively, one may suspect that the absence of temporal isolation conflict between the processes of “integration” and “separation” across
effects in the spatial-continuous condition might result from forward space and time. Strictly speaking, our results did not rule out the pos­
serial retrieval of the encoded information given the continuous nature sibility that the effects of spatial integration and temporal isolation
of movement sequences. As shown in the previous research, temporal might occur simultaneously. But even so, the null effect of ISI in the
isolation did not benefit memory performance in forward serial recall (e. spatial-continuous condition would suggest that the performance
g., Lewandowsky et al., 2006; Nimmo & Lewandowsky, 2005; Par­ enhancement due to temporal isolation was fully offset against the
mentier et al., 2006), as the temporal information is less relevant to diminution of integration benefits given decreased temporal proximity,
memory retrieval under this condition (Geiger & Lewandowsky, 2008; and this would again indicate a trade-off between spatial integration and
Lewandowsky et al., 2008). However, this possibility cannot explain the temporal isolation.
current finding. For one thing, the single-probe recognition task, by
design, did not require order information to be retrieved; for another 3. Experiment 2
thing, the one-item recency effect (as shown in accuracy and RTs, see
Section 2.2.2) is typical of item recognition task as opposed to a serial In Experiment 1, we found that the continuity state of a movement
position curve with both primacy and recency effects normally obtained sequence could influence working memory for its constituent units.
in sequential recall (see Oberauer et al., 2018 for a review). The results However, as discussed previously, spatiotemporal continuity of move­
therefore suggest that participants retrieved the probed movement ments may facilitate integration through bottom-up Gestalt principles or
directly from memory rather than replaying the whole sequence from the top-down predictive processing. It is therefore unclear whether a
beginning and scanning for a match item-by-item (i.e., forward serial stimulus-driven mechanism is sufficient to induce the observed effects
retrieval). Note that with the word “directly”, we emphasized the time from spatiotemporal continuity or a prediction-based mechanism might
course of the retrieval process — whether a serial comparison based on be required. To distinguish between these two mechanisms — one is
input order was involved — rather than how the observed information driven by bottom-up perceptual properties of a movement sequence and
was represented in memory. In other words, item-based retrieval did not the other is mediated by the subjective prediction about the continuity
reject the possibility of integrative processing during perception or state, we abolished the predictability of the continuity state in Experi­
memory encoding. For example, the probed movement might be ment 2 and examined how this might influence the presence or absence
compared with (or fit into) an integrated representation of the sample of the observed effects in Experiment 1. Specifically, if a prediction-
sequence, such as a configuration of movement trajectory, or it might be based mechanism is required, the unpredictability of the continuity
compared with item information further decoded from such higher- state would diminish benefits from both spatial integration and tem­
order representations. poral isolation, but if a stimulus-driven mechanism is sufficient to
Moreover, it is worth noting that the starting posture of movement induce integration and separation processes, respectively, both effects
units might also influence recognition performance. Specifically, would still be observable.

5
S.-C. Chiou Acta Psychologica 230 (2022) 103719

According to the finding in Experiment 1, an observer can fill the Fig. 5). All sequences were displayed with ISIs of 0 s, 0.6 s, 1.2 s, or 1.8 s,
temporal gaps to benefit from perceptual integration if the succeeding each accounting for one fourth of the total trials. The longest ISI was
unit is to form a continuous trajectory with the current one; otherwise, extended from 1.5 s (in Experiment 1) to 1.8 s, which was assumed to
the observer can leave the gaps “unfilled” to benefit from temporal better induce temporal isolation effects, if any. Participants completed 8
isolation. Since these two processes contradict each other, a prediction- practice trials and then 240 fully-randomized experimental trials (with
based mechanism might be required to determine which strategy better 50 % positive trials) in eight blocks. Other procedures remained the
fits the ongoing situation. Moreover, due to the sequential display of same as in Experiment 1.
movement units, correct classification of movement sequences (contin­
uous vs. discontinuous) as well as correct selection of processing stra­ 3.1.3. Data analysis
tegies (integration vs. separation) might not be possible before the If the manipulation was successful (i.e., if the continuity state of
display of the succeeding unit if no prior knowledge is available. As a movement sequences was unpredictable), the bottom-up characteristics
result, we predicted that a prediction-based mechanism might be of movement sequences would be the most useful information for par­
involved in the process and thus the unpredictability of the continuity ticipants to categorize movement units and to benefit from potential
state would diminish benefits from both spatial integration and tem­ effects of spatiotemporal (dis)continuity. Specifically, participants may
poral isolation. treat a movement unit as continuous (or discontinuous) if it was pre­
ceded or followed by another unit with which a continuous (or a
discontinuous) trajectory can be formed. However, except for fully
3.1. Method
continuous (1234) or discontinuous (5768) sequences, it was ambiguous
whether a specific unit within a “hybrid” sequence would be regarded as
3.1.1. Participants
continuous or discontinuous. For example, the third unit of the sequence
The same sample size as in Experiment 1 (N = 24) was originally
type 5823 might be integrated with the last unit to construct a contin­
chosen to detect the same effects of spatial integration and temporal
uous trajectory (23), or it might be processed with the first two units as
isolation, if any, under a condition in which the continuity state of
part of a discontinuous sequence (582). Due to this ambiguity, move­
movement sequences was unpredictable. One participant was excluded
ment units that might benefit from spatial integration (i.e., those being
from analyses due to poor performance (out of 2 standard deviations
treated as continuous) and movement units that might enjoy temporal
from the group mean), leaving a final sample of 23 participants (15 fe­
isolation (i.e., those being treated as discontinuous) cannot be clearly
male; aged 18–31 years, M = 23.8, SD = 3.5). No participants have
distinguished.
viewed the experimental materials before. Participants' expertise in­
To solve this problem, data were analyzed under five independent
dexes (0–4) in dance, music, and sport were 0.2 (SD = 0.4), 0.9 (SD =
assumptions concerning how strong this bottom-up tendency might be:
0.9), and 1.3 (SD = 1.0), respectively, indicating limited expertise. No
I. Four-unit Continuity, in which only movement units that form a four-
significant difference was shown between participants' expertise indexes
unit continuous (or discontinuous) sequence were treated as contin­
in Experiment 1 and 2, all ps ≥ 0.217. All participants signed informed
uous (or discontinuous), such as 2 in 1234, II. Three-unit Continuity, in
consent prior to the experiment and received €8 per hour for their
which movement units that form a three-unit (or above) continuous (or
participation.
discontinuous) sequence were treated as continuous (or discontinuous),
such as 2 in 1238, III. Two-unit Continuity, in which movement units that
3.1.2. Stimuli and procedure
form a two-unit (or above) continuous (or discontinuous) sequence were
The same thirty movement sequences were used as in Experiment 1.
treated as continuous (or discontinuous), such as 2 in 1267, except those
To abolish the predictability of the continuity state, we altered the
being part of a longer discontinuous (or continuous) sequence, such as 2
original block design into a completely randomized design. In addition,
in 1287, and those surrounded by two discontinuous (or continuous)
we reordered the movement units and equalized the chance of pre­
sequences, such as 3 in 5346, IV. Continuity as Default, in which partic­
senting a continuous or a discontinuous succeeding unit at three tran­
ipants treated all movement units that could possibly form a continuous
sition points of a four-unit sequence, yielding eight possible sequence
sequence as continuous, irrespective of whether this unit could also be
types in total (see Fig. 5). With this sequence design, participants could
perceived as part of a discontinuous sequence, such as 2 in 1287 and 3 in
no longer know in advance whether the upcoming unit would form a
5346, and V. Discontinuity as Default, in which participants treated all
continuous or a discontinuous trajectory with the current one. The new
movement units that could possibly form a discontinuous sequence as
sequences were composed of four movement units from the same orig­
discontinuous, such as 3 in 1238 and 2 in 1267. Fig. 6 illustrates
inal sequence (i.e., sequences 1234 and 5678) or from two different
movement units that were included into analysis under these five
sequences (i.e., a recombination of the sequences 1234 and 5678) (see

Fig. 5. Sequence design in Experiment 2. The predictability of the continuity state was abolished through an equal chance of presenting a continuous or a
discontinuous succeeding unit at three transition points of a four-unit sequence. Eight possible sequence types (a to h) were composed of four movement units from
the same original sequence (a: 1234 and h: 5768) or from two different sequences (b:1287, c: 5823, d: 5346, e: 1238, f: 5234, and g: 1267).

6
S.-C. Chiou Acta Psychologica 230 (2022) 103719

Fig. 6. Memory performance (measured by Hits – FAs) in Experiment 2 analyzed under five independent assumptions concerning how participants might process the
observed information based on bottom-up characteristics of movement sequences: I. Four-unit Continuity, in which only movement units that form a four-unit
continuous (or discontinuous) sequence were treated as continuous (or discontinuous), such as 2 in 1234, II. Three-unit Continuity, in which movement units that
form a three-unit (or above) continuous (or discontinuous) sequence were treated as continuous (or discontinuous), such as 2 in 1238, III. Two-unit Continuity, in
which movement units that form a two-unit (or above) continuous (or discontinuous) sequence were treated as continuous (or discontinuous), such as 2 in 1267,
except those being part of a longer discontinuous (or continuous) sequence, such as 2 in 1287, and those surrounded by two discontinuous (or continuous) sequences,
such as 3 in 5346, IV. Continuity as Default, in which participants treated all movement units that could possibly form a continuous sequence as continuous, irre­
spective of whether this unit could also be perceived as part of a discontinuous sequence, such as 2 in 1287 and 3 in 5346, and V. Discontinuity as Default, in which
participants treated all movement units that could possibly form a discontinuous sequence as discontinuous, such as 3 in 1238 and 2 in 1267. Colored squares
illustrate movement units that were included into analysis under these five assumptions, respectively. Error bars indicate 95 % confidence intervals calculated across
participants.

assumptions. 3.2. Results


Same as Experiment 1, the data were submitted to a repeated-
measures ANOVA to test whether there was any effect from Spatial 3.2.1. Overall performance: Hits – FAs
Continuity (i.e., spatial integration) or ISI (i.e., temporal isolation). To Five 2 (Spatial Continuity: continuous, discontinuous) × 4 (ISI: 0 s,
support our hypothesis that a prediction-based mechanism might be 0.6 s, 1.2 s, 1.8 s) repeated-measures ANOVAs were conducted on Hits –
involved in the process, neither spatial integration nor temporal isola­ FAs under the aforementioned five assumptions, respectively. The
tion effects should be observed under all these assumptions. Note that complete ANOVA results are summarized in Table 1 and illustrated in
the five assumptions we made for data analysis were conditions in which Fig. 6. Across all analyses, the main effect for Spatial Continuity was not
the effect of spatiotemporal (dis)continuity was most likely to be significant, 0.426 ≤ ps ≤ 0.953. Bayesian analysis also provided sub­
observed (and thus with the highest chance to reject our hypothesis). It stantial evidence in support of this null effect, 4.95 ≤ BF01 (Bayes Factor
is also possible, for example, that participants totally failed to categorize in favor of the null hypothesis) ≤ 6.45. The smallest BF01 of 4.95 indi­
movement units even with the presence of bottom-up information. cated that the data was at least about five times more likely under a
Although this scenario was not included in the analysis, the expected model assuming no genuine effect from Spatial Continuity. The results
null effect would still support the hypothesis, even in a stronger way, implied that movement units that formed a continuous trajectory were
that a prediction-based mechanism might be involved. Moreover, as a not memorized better than those that formed a discontinuous trajectory,
conclusion may need to be drawn from null effects, we adopted Bayes i.e., no spatial integration benefits. The main effect for ISI was not sig­
Factors (Kass & Raftery, 1995) as an alternative inferential statistic and nificant, either, 0.067 ≤ ps ≤ 0.478, indicating that no temporal isola­
conducted the analyses by using statistical software program JASP tion effects were obtained for movement units displayed with temporal
(version 0.11.1; JASP Team, 2019) with the default multivariate Cauchy gaps. Note that although the evidence from Bayesian analysis supported
priors (scale parameter = 0.5 for fixed effects and 1 for random effects). the null hypothesis in general (i.e., all BF01 > 1), it provided only

7
S.-C. Chiou Acta Psychologica 230 (2022) 103719

Table 1 memory rather than retrieving the whole sequence and scanning for a
Results of the five 2 spatial continuity (CONT) × 4 inter-stimulus interval (ISI) match sequentially. The finding is consistent with Experiment 1.
repeated-measures ANOVAs in Experiment 2.
Factors df F p η2p BF10 BF01
3.3. Discussion
I. Four-unit Continuity
CONT 1, 22 0.324 0.575 0.015 0.171 5.848
In Experiment 2, we examined whether a prediction-based mecha­
ISI 3, 66 0.837 0.478 0.037 0.078 12.821
CONT × ISI 3, 66 2.371 0.078 0.097 1.077 0.929 nism might contribute to the effects of spatiotemporal (dis)continuity
when observing and memorizing whole-body movement sequences. We
found that neither spatial integration nor temporal isolation effects were
II. Three-unit continuity
CONT 1, 22 0.602 0.446 0.027 0.190 5.263 present when the continuity state of movement sequences was unpre­
ISI 3, 66 1.884 0.141 0.079 0.260 3.846 dictable, indicating that a prediction-based mechanism might be
CONT × ISI 3, 66 0.493 0.688 0.022 0.096 10.400 involved in the process. Nevertheless, there was an upward trend of
performance, albeit non-significant, for both spatial-continuous and
III. Two-unit continuity -discontinuous sequences as ISIs increased (see Fig. 6). The finding in­
CONT 1, 22 0.659 0.426 0.029 0.202 4.950 dicates that there might be a tendency in favor of separation process and
ISI 3, 66 2.280 0.087 0.094 0.767 1.304
that the temporal gaps might still benefit memory performance despite
CONT × ISI 3, 66 0.999 0.399 0.043 0.144 6.955
the unpredictability of the continuity state.

IV. Continuity as default


CONT 1, 22 0.004 0.953 0.000 0.155 6.452 4. General discussion
ISI 3, 66 1.178 0.325 0.051 0.120 8.333
CONT × ISI 3, 66 0.825 0.485 0.036 0.150 6.667 The present research investigated two questions: First, how the
spatiotemporal (dis)continuity of a movement sequence influences
V. Discontinuity as default working memory for its constituent units (Experiment 1); second,
CONT 1, 22 0.235 0.633 0.011 0.180 5.556 whether a bottom-up, stimulus-driven mechanism is sufficient to induce
ISI 3, 66 2.505 0.067 0.102 0.530 1.887
the effects of spatiotemporal (dis)continuity when observing and
CONT × ISI 3, 66 0.345 0.793 0.015 0.087 11.500
memorizing a movement sequence, or a prediction-based mechanism
might be required (Experiment 2).
anecdotal evidence for this null effect (i.e., BF01 < 3) under two of the We found that the contextual information of a movement — whether
assumptions (i.e., assumption III and V, see Table 1). Collapsing all it is part of a continuous or a discontinuous sequence — affected
conditions, Cowan's (2001) formula showed that participants were able working memory for this movement. The finding suggests that partici­
to retain 2.89 movement units in working memory, consistent with the pants did not simply process the information of individual movements,
finding of Experiment 1 and previous research (Shen et al., 2014; Wood, but also how these movements were combined with one another in a
2007). sequence. Previous studies have shown that when participants were
asked to memorize a sequence of spatial locations, the complexity of the
3.2.2. Serial-position effects: accuracy (hit rate) and RTs for positive trials to-be-remembered sequences, as measured by path crossing, affected
Same as Experiment 1, we conducted a one-way repeated-measures performance of serial recall (Parmentier et al., 2005; Parmentier &
ANOVA on accuracy (hit rate) and RTs for positive trials to examine the Andrés, 2006). The finding indicates that participants processed loca­
influence of input serial position on memory performance. The analysis tion information as well as “transition” information and that spatial
on hit rate yielded a significant main effect for serial position, F(3, 66) = characteristics of the sequences played a pivotal role in visuospatial
14.3, p < .001, η2p = 0.40. Post hoc analyses indicated a recency effect, ps serial memory. Recent studies also showed that contextual information
≤ 0.002, but no primacy effect, p = .530 (Fig. 7a). Similar results were could affect item memory even in a very simple, unstructured display (e.
obtained from the analysis on RTs, F(3, 66) = 38.4, p < .001, η2p = 0.64. g., Liesefeld et al., 2019), illustrating the importance of considering
The shortest RTs occurred when the recognition probe was the last unit global-level representations in working memory research. Note that,
of a sequence, ps < 0.001, while RTs for the first three units did not show however, the current study did not require any contextual or order in­
significant difference, ps ≥ 0.355 (Fig. 7b). The one-item recency effect formation to be retrieved. The one-item recency effect, which is opposed
suggests that participants retrieved the probed movement directly from to a serial position curve with both primacy and recency effects normally
obtained from sequential recall (Oberauer et al., 2018), provided

Fig. 7. Serial-position effects in Experiment 2 (after collapsing ISIs) when measured by hit rate (a) and response times (RTs) (b). Spatial-continuous and -discon­
tinuous units were classified here based on the assumption III. Two-unit Continuity, in which all movement units were included into analysis and equally distributed
into continuous and discontinuous categories (see Section 3.1.3 and Fig. 6). Error bars indicate 95 % confidence intervals calculated across participants.

8
S.-C. Chiou Acta Psychologica 230 (2022) 103719

evidence for the item-based retrieval in both experiments. Therefore, the finding suggests that participants might have treated all movement units
observed effects from spatiotemporal (dis)continuity might occur in as discontinuous irrespective of whether they were continuous or
earlier stages of information processing such as when a movement discontinuous. As no prior knowledge about the continuity state was
sequence was perceived and encoded. available, we suspect that this tendency in favor of separation process
In addition, we found a stronger effect from integration over sepa­ might reflect the bottom-up influence from stimulus properties. For
ration in the spatial domain (spatial integration) given temporal conti­ example, temporal gaps (present in 75 % of the trials), which were
nuity, and a larger influence from separation over integration in the shown as black screens inserted between movement units, might serve
temporal domain (temporal isolation) given spatial discontinuity. More as natural segmentations due to the resulting perceptual discontinuity (i.
importantly, our results demonstrated a trade-off between spatial inte­ e., discontinuity of video display). In addition, since the spatial conti­
gration and temporal isolation, indicating a conflict between the pro­ nuity (i.e., continuity of movement trajectory) was defined in its loosest
cesses of integration and separation across space and time. Specifically, form with the discrete (self-sufficient) nature of individual movement
when temporal gaps were present, participants could choose to fill the units being preserved (see Section 2.1.2), this might also encourage
gaps by prolonging mental representations of individual movement separation process. Alternatively, since we did not manipulate the pre­
units (integration process), or make use of the gaps to form effective dictability of ISIs within trials (i.e., temporal gaps were either present or
isolation (separation process). However, they cannot benefit from the absent in a trial), participants might be able to accommodate their
both at the same time, as illustrated by the absence of temporal isolation processing strategies after seeing the first two units of a sequence, i.e.,
effects in the spatial-continuous condition (Experiment 1) and the after knowing whether there would be temporal gaps in the remaining
absence of integration benefits when spatial-continuous sequences could part of the trial and whether it might be possible to benefit from tem­
possibly enjoy temporal isolation (Experiment 2). poral isolation. As the continuity state of movement sequences was
Nevertheless, there was a processing advantage in the spatial- unpredictable, a simple strategy would be to treat all movement units as
continuous condition over the spatial-discontinuous condition when a discontinuous if temporal gaps were present irrespective of whether the
wide range of temporal gaps were applied. The finding indicates that a movement units were indeed discontinuous.
decrease of temporal proximity did not impair positive effects from Furthermore, although the key manipulation of the current study —
spatial continuity and that the processes beyond low-level motion spatiotemporal continuity of an observed movement sequence — was
analysis might be involved. In fact, it has long been suggested that top- purely in visual modality, the advantage we found in processing
down mechanisms, such as configural processing of human body form continuous movement sequences might also be attributed to kin­
(Orgs et al., 2011), categorical processing of human action (Dittrich, aesthetic factors. Research on mirror-neuron system (Cook et al., 2014;
1993), and focus of attention (Cavanagh et al., 2001; Thornton et al., Gallese & Goldman, 1998; Rizzolatti & Craighero, 2004) suggests covert
2002), play a role in human motion perception (see Thornton, 2013 for a motor simulation during action observation; studies on biological
review). In the current study, we further demonstrated that a prediction- apparent motion also showed recruitment of motor areas in constructing
based mechanism (i.e., prior knowledge or belief) might be involved in fluent movement perception from static body postures (Orgs et al.,
the perception of whole-body movement sequences and the spatiotem­ 2016). It is thus reasonable to speculate that a continuous movement
poral integration of movement information. Specifically, participants display would enable a smoother motor simulation, if any, and assist in
constructed a continuous movement representation only if they knew working memory encoding by strengthening the otherwise purely
that a continuous sequence would be displayed. When the continuity visual-based memory traces. On the contrary, an “interrupted” simula­
state was unpredictable, no integration benefits were present even when tion due to a discontinuous movement display might create conflicts
no gaps were to be filled (i.e., when ISI = 0). The finding also resonates from the perspective of motor control. Moreover, the physical knowl­
with the basic assumption of the event segmentation theory (Zacks et al., edge that participants have acquired from their lifelong motor experi­
2007), which suggests that observers continuously make predictions ence might also afford an efficient encoding strategy, such as by
about the upcoming events, grouping predictable perceptual experience restricting the possibilities of succeeding movements based on the
into a gestalt or dividing it into parts when transient prediction errors intrinsic, spatiotemporal-continuous nature of human movements.
arise (Zacks, 2020). Further studies are required to delineate the sensorimotor contribution
On the contrary, a decrease of temporal proximity reversely to the effects of spatiotemporal (dis)continuity observed in the current
enhanced memory performance for spatial-discontinuous sequences, study.
consistent with the prediction of the temporal distinctiveness account of In conclusion, our results showed an influence from the global-level
working memory (Brown et al., 2000; Brown et al., 2007; Glenberg & contextual information (i.e., spatiotemporal continuity) on working
Swanson, 1986; Neath & Crowder, 1996). However, as discussed pre­ memory for movements. More importantly, we demonstrated that the
viously, it is also possible that temporal gaps simply provided extra time seemingly stimulus-driven perceptual effects from spatiotemporal con­
for memory consolidation (Lewandowsky & Brown, 2005; Ricker & tinuity might be supported by a prediction-based mechanism. Further­
Cowan, 2014) or allowed encoding resource to be replenished (Mizrak & more, a trade-off we found between spatial integration and temporal
Oberauer, 2021; Popov & Reder, 2020) rather than facilitating distinc­ isolation illustrates an interweaving relationship between spatial and
tive processing. Although these hypotheses seem equally plausible to temporal processing during action observation, and highlights the
explain the current finding, a trade-off we found between spatial inte­ importance of considering the dynamic and continuous nature of human
gration and temporal isolation in the spatial-continuous condition pro­ movements in visual perception and working memory research.
vided indirect evidence in support of the distinctiveness account.
Specifically, a continuous representation of movement trajectory would Declaration of competing interest
reduce effective isolation, but not the available time for memory
consolidation or resource recovery. Therefore, the distinctiveness ac­ The author has no conflict of interest to declare.
count would predict no isolation effects when memorizing spatial-
continuous sequences, while the beneficial effect of time should still Acknowledgements
be observable according to the consolidation or resource recovery ac­
count. Since no isolation effects were obtained in the spatial-continuous This research was supported by the Cluster of Excellence Cognitive
condition, the distinctiveness account was better supported. Interaction Technology ‘CITEC’ (EXC 277) at Bielefeld University, which
It is worth noting that there was a modest temporal isolation effect, has been funded by the German Research Foundation (DFG). The author
albeit non-significant, for both continuous and discontinuous sequences thanks Thomas Schack for his helpful remarks on the manuscript and his
when the continuity state was unpredictable (Experiment 2). The generous support to the research project. The author also thanks Theater

9
S.-C. Chiou Acta Psychologica 230 (2022) 103719

Bielefeld for supporting the movement recording, and Tiemen Ste­ Lewandowsky, S., & Brown, G. D. (2005). Serial recall and presentation schedule: A
micro-analysis of local distinctiveness. Memory, 13(3–4), 283–292. https://doi.org/
merding and Inês Carijó for demonstrating movement sequences. The
10.1080/09658210344000251
author acknowledges support for the publication costs by the Open Lewandowsky, S., Brown, G. D., Wright, T., & Nimmo, L. M. (2006). Timeless memory:
Access Publication Fund of Bielefeld University. Evidence against temporal distinctiveness models of short-term memory for serial
order. Journal of Memory and Language, 54(1), 20–38. https://doi.org/10.1016/j.
jml.2005.08.004
Appendix A. Supplementary data Lewandowsky, S., Nimmo, L. M., & Brown, G. D. (2008). When temporal isolation
benefits memory for serial order. Journal of Memory and Language, 58(2), 415–428.
Supplementary data to this article can be found online at https://doi. https://doi.org/10.1016/j.jml.2006.11.003
Liesefeld, H. R., Liesefeld, A. M., & Müller, H. J. (2019). Two good reasons to say
org/10.1016/j.actpsy.2022.103719. ‘change!’–Ensemble representations as well as item representations impact standard
measures of VWM capacity. British Journal of Psychology, 110(2), 328–356. https://
References doi.org/10.1111/bjop.12359
Miller, G. A. (1956). The magical number seven, plus or minus two: Some limits on our
capacity for processing information. Psychological Review, 63(2), 81–97. https://doi.
Abend, W., Bizzi, E., & Morasso, P. (1982). Human arm trajectory formation. Brain, 105
org/10.1037/h0043158
(Pt 2), 331–348. https://doi.org/10.1093/brain/105.2.331
Mizrak, E., & Oberauer, K. (2021). What is time good for in working memory?
Bertenthal, B. I., & Pinto, J. (1994). Global processing of biological motions. Psychological
Psychological Science, 32(8), 1325–1337. https://doi.org/10.1177/
Science, 5(4), 221–225. https://doi.org/10.1111/j.1467-9280.1994.tb00504.x
0956797621996659
Brady, T. F., Konkle, T., & Alvarez, G. A. (2009). Compression in visual working memory:
Morin, C., Brown, G. D., & Lewandowsky, S. (2010). Temporal isolation effects in
Using statistical regularities to form more efficient memory representations. Journal
recognition and serial recall. Memory & Cognition, 38(7), 849–859. https://doi.org/
of Experimental Psychology: General, 138(4), 487–502. https://doi.org/10.1037/
10.3758/MC.38.7.849
a0016797
Nassar, M. R., Helmers, J. C., & Frank, M. J. (2018). Chunking as a rational strategy for
Brady, T. F., Konkle, T., & Alvarez, G. A. (2011). A review of visual memory capacity:
lossy data compression in visual working memory. Psychological Review, 125(4),
Beyond individual items and toward structured representations. Journal of Vision, 11
486–511. https://doi.org/10.1037/rev0000101
(5). https://doi.org/10.1167/11.5.4
Neath, I., & Crowder, R. G. (1996). Distinctiveness and very short-term serial position
Brown, G. D., Morin, C., & Lewandowsky, S. (2006). Evidence for time-based models of
effects. Memory, 4(3), 225–242. https://doi.org/10.1080/09658211.1996.9753032
free recall. Psychonomic Bulletin & Review, 13(4), 717–723. https://doi.org/10.3758/
Nimmo, L. M., & Lewandowsky, S. (2005). From brief gaps to very long pauses: Temporal
bf03193986
isolation does not benefit serial recall. Psychonomic Bulletin & Review, 12(6),
Brown, G. D., Neath, I., & Chater, N. (2007). A temporal ratio model of memory.
999–1004.
Psychological Review, 114(3), 539–576. https://doi.org/10.1037/0033-
Oberauer, K., Lewandowsky, S., Awh, E., Brown, G. D. A., Conway, A., Cowan, N.,
295X.114.3.539
Donkin, C., Farrell, S., Hitch, G. J., Hurlstone, M. J., Ma, W. J., Morey, C. C.,
Brown, G. D., Preece, T., & Hulme, C. (2000). Oscillator-based memory for serial order.
Nee, D. E., Schweppe, J., Vergauwe, E., & Ward, G. (2018). Benchmarks for models
Psychological Review, 107(1), 127–181. https://doi.org/10.1037/0033-
of short term and working memory. Psychological Bulletin, 144(9), 885–958. https://
295x.107.1.127
doi.org/10.1037/bul0000153
Cavanagh, P., Labianca, A. T., & Thornton, I. M. (2001). Attention-based visual routines:
Orgs, G., Bestmann, S., Schuur, F., & Haggard, P. (2011). From body form to biological
Sprites. Cognition, 80(1–2), 47–60. https://doi.org/10.1016/s0010-0277(00)00153-
motion: The apparent velocity of human movement biases subjective time.
0
Psychological Science, 22(6), 712–717. https://doi.org/10.1177/0956797611406446
Chatterjee, S. H., Freyd, J. J., & Shiffrar, M. (1996). Configural processing in the
Orgs, G., Dovern, A., Hagura, N., Haggard, P., Fink, G. R., & Weiss, P. H. (2016).
perception of apparent biological motion. Journal of Experimental Psychology: Human
Constructing visual perception of body movement with the motor cortex. Cerebral
Perception and Performance, 22(4), 916–929. https://doi.org/10.1037//0096-
Cortex, 26(1), 440–449. https://doi.org/10.1093/cercor/bhv262
1523.22.4.916
Palmer, S., & Rock, I. (1994). Rethinking perceptual organization: The role of uniform
Chiou, S.-C. (2022). Attention modulates incidental memory encoding of human
connectedness. Psychonomic Bulletin & Review, 1(1), 29–55. https://doi.org/
movements. Cognitive Processing, 23(2), 155–168. https://doi.org/10.1007/s10339-
10.3758/bf03200760
022-01078-1
Parmentier, F. B., & Andrés, P. (2006). The impact of path crossing on visuo-spatial serial
Cook, R., Bird, G., Catmur, C., Press, C., & Heyes, C. (2014). Mirror neurons: From origin
memory: Encoding or rehearsal effect? Quarterly Journal of Experimental Psychology,
to function. Behavioral and Brain Sciences, 37(2), 177–192. https://doi.org/10.1017/
59(11), 1867–1874. https://doi.org/10.1080/17470210600872154
S0140525X13000903
Parmentier, F. B., Elford, G., & Maybery, M. (2005). Transitional information in spatial
Cowan, N. (2001). The magical number 4 in short-term memory: A reconsideration of
serial memory: Path characteristics affect recall performance. Journal of Experimental
mental storage capacity. Behavioral and Brain Sciences, 24(1), 87–114. https://doi.
Psychology: Learning, Memory, and Cognition, 31(3), 412–427. https://doi.org/
org/10.1017/s0140525x01003922
10.1037/0278-7393.31.3.412
Cowan, N., Chen, Z., & Rouder, J. N. (2004). Constant capacity in an immediate serial-
Parmentier, F. B., King, S., & Dennis, I. (2006). Local temporal distinctiveness does not
recall task: A logical sequel to Miller (1956). Psychological Science, 15(9), 634–640.
benefit auditory verbal and spatial serial recall. Psychonomic Bulletin & Review, 13(3),
https://doi.org/10.1111/j.0956-7976.2004.00732.x
458–465. https://doi.org/10.3758/BF03193870
Dittrich, W. H. (1993). Action categories and the perception of biological motion.
Peirce, J. W. (2007). PsychoPy–Psychophysics software in python. Journal of
Perception, 22(1), 15–22. https://doi.org/10.1068/p220015
Neuroscience Methods, 162(1–2), 8–13. https://doi.org/10.1016/j.
Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007). G* Power 3: A flexible
jneumeth.2006.11.017
statistical power analysis program for the social, behavioral, and biomedical
Peirce, J. W. (2009). Generating stimuli for neuroscience using PsychoPy. Frontiers in
sciences. Behavior Research Methods, 39(2), 175–191. https://doi.org/10.3758/
Neuroinformatics, 2(10), 10. https://doi.org/10.3389/neuro.11.010.2008
BF03193146
Pettijohn, K. A., Thompson, A. N., Tamplin, A. K., Krawietz, S. A., & Radvansky, G. A.
Freyd, J. J., & Finke, R. A. (1984). Representational momentum. Journal of Experimental
(2016). Event boundaries and memory improvement. Cognition, 148, 136–144.
Psychology: Learning, Memory, and Cognition, 10(1), 126–132. https://doi.org/
https://doi.org/10.1016/j.cognition.2015.12.013
10.1037/0278-7393.10.1.126
Popov, V., & Reder, L. M. (2020). Frequency effects on memory: A resource-limited
Gallese, V., & Goldman, A. (1998). Mirror neurons and the simulation theory of mind-
theory. Psychological Review, 127(1), 1–46. https://doi.org/10.1037/rev0000161
reading. Trends in Cognitive Sciences, 2(12), 493–501. https://doi.org/10.1016/
Richmond, L. L., & Zacks, J. M. (2017). Constructing experience: Event models from
s1364-6613(98)01262-5
perception to action. Trends in Cognitive Sciences, 21(12), 962–980. https://doi.org/
Geiger, S. M., & Lewandowsky, S. (2008). Temporal isolation does not facilitate forward
10.1016/j.tics.2017.08.005
serial recall—Or does it? Memory & Cognition, 36(5), 957–967. https://doi.org/
Ricker, T. J., & Cowan, N. (2014). Differences between presentation methods in working
10.3758/MC.36.5.957
memory procedures: A matter of working memory consolidation. Journal of
Glenberg, A. M., & Swanson, N. G. (1986). A temporal distinctiveness theory of recency
Experimental Psychology: Learning, Memory, and Cognition, 40(2), 417–428. https://
and modality effects. Journal of Experimental Psychology: Learning, Memory, and
doi.org/10.1037/a0034301
Cognition, 12(1), 3–15. https://doi.org/10.1037//0278-7393.12.1.3
Rizzolatti, G., & Craighero, L. (2004). The mirror-neuron system. Annual Review of
Hubbard, T. L. (2005). Representational momentum and related displacements in spatial
Neuroscience, 27(1), 169–192. https://doi.org/10.1146/annurev.
memory: A review of the findings. Psychonomic Bulletin & Review, 12(5), 822–851.
neuro.27.070203.144230
https://doi.org/10.3758/BF03196775
Rouder, J. N., Morey, R. D., Morey, C. C., & Cowan, N. (2011). How to measure working
Hunt, R. R., & Worthen, J. B. (2006). Distinctiveness and memory. Oxford University Press.
memory capacity in the change detection paradigm. Psychonomic Bulletin & Review,
JASP Team. (2019). JASP (Version 0.11.1)[Computer software]. https://jasp-stats.org/.
18(2), 324–330. https://doi.org/10.3758/s13423-011-0055-3
Johansson, G. (1973). Visual perception of biological motion and a model for its analysis.
Shen, M., Gao, Z., Ding, X., Zhou, B., & Huang, X. (2014). Holding biological motion
Perception & Psychophysics, 14(2), 201–211. https://doi.org/10.3758/BF03212378
information in working memory. Journal of Experimental Psychology: Human
Johansson, G. (1976). Spatio-temporal differentiation and integration in visual motion
Perception and Performance, 40(4), 1332–1345. https://doi.org/10.1037/a0036839
perception. Psychological Research, 38(4), 379–393. https://doi.org/10.1007/
Shiffrar, M., & Freyd, J. J. (1990). Apparent motion of the human body. Psychological
BF00309043
Science, 1(4), 257–264. https://doi.org/10.1111/j.1467-9280.1990.tb00210.x
Kass, R. E., & Raftery, A. E. (1995). Bayes factors. Journal of the American Statistical
Shiffrar, M., & Freyd, J. J. (1993). Timing and apparent motion path choice with human
Association, 90(430), 773–795. https://doi.org/10.1080/01621459.1995.10476572
body photographs. Psychological Science, 4(6), 379–384. https://doi.org/10.1111/
j.1467-9280.1993.tb00585.x

10
S.-C. Chiou Acta Psychologica 230 (2022) 103719

Shiffrar, M., Lichtey, L., & Chatterjee, S. H. (1997). The perception of biological motion Perceptual grouping and figure–ground organization. Psychological Bulletin, 138(6),
across apertures. Perception & Psychophysics, 59(1), 51–59. https://doi.org/10.3758/ 1172–1217. https://doi.org/10.1037/a0029333
BF03206847 Wertheimer, M. (1923). Untersuchungen zur Lehre von der Gestalt. II [Investigations in
Sternberg, S. (1969). Memory-scanning: Mental processes revealed by reaction-time Gestalt Theory: II. Laws of organization in perceptual forms]. Psychologische
experiments. American Scientist, 57(4), 421–457. Forschung, 4(1), 301–350. https://doi.org/10.1007/BF00410640
Thornton, I. M. (2013). Top-down versus bottom-up processing of biological motion. In Wood, J. N. (2007). Visual working memory for observed actions. Journal of Experimental
People watching: Social, perceptual, and neurophysiological studies of body perception Psychology: General, 136(4), 639–652. https://doi.org/10.1037/0096-
(pp. 25–43). Oxford University Press. 3445.136.4.639
Thornton, I. M., Pinto, J., & Shiffrar, M. (1998). The visual perception of human Xu, Y., & Chun, M. M. (2007). Visual grouping in human parietal cortex. Proceedings of
locomotion. Cognitive Neuropsychology, 15(6–8), 535–552. https://doi.org/10.1080/ the National Academy of Sciences, 104(47), 18766–18771. https://doi.org/10.1073/
026432998381014 pnas.0705618104
Thornton, I. M., Rensink, R. A., & Shiffrar, M. (2002). Active versus passive processing of Zacks, J. M. (2020). Event perception and memory. Annual Review of Psychology, 71(1),
biological motion. Perception, 31(7), 837–853. https://doi.org/10.1068/p3072 165–191. https://doi.org/10.1146/annurev-psych-010419-051101
Wagemans, J., Elder, J. H., Kubovy, M., Palmer, S. E., Peterson, M. A., Singh, M., & von Zacks, J. M., Speer, N. K., Swallow, K. M., Braver, T. S., & Reynolds, J. R. (2007). Event
der Heydt, R. (2012). A century of Gestalt psychology in visual perception: I. perception: A mind-brain perspective. Psychological Bulletin, 133(2), 273–293.
https://doi.org/10.1037/0033-2909.133.2.273

11

You might also like