Research Article: The Frozen Effect: Objects in motion are more aesthetically appealing than objects frozen in time

Date Published: May 16, 2019

Publisher: Public Library of Science

Author(s): Malerie G. McDowell, Jason Haberman, Cosimo Urgesi.


Videos of moving faces are more flattering than static images of the same face, a phenomenon dubbed the Frozen Face Effect. This may reflect an aesthetic preference for faces viewed in a more ecological context than still photographs. In the current set of experiments, we sought to determine whether this effect is unique to facial processing, or if motion confers an aesthetic benefit to other stimulus categories as well, such as bodies and objects—that is, a more generalized ‘Frozen Effect’ (FE). If motion were the critical factor in the FE, we would expect the video of a body or object in motion to be significantly more appealing than when seen in individual, static frames. To examine this, we asked participants to rate sets of videos of bodies and objects in motion along with the still frames constituting each video. Extending the original FFE, we found that participants rated videos as significantly more flattering than each video’s corresponding still images, regardless of stimulus domain, suggesting that the FFE generalizes well beyond face perception. Interestingly, the magnitude of the FE increased with the predictability of stimulus movement. Our results suggest that observers prefer bodies and objects in motion over the same information presented in static form, and the more predictable the motion, the stronger the preference. Motion imbues objects and bodies with greater aesthetic appeal, which has implications for how one might choose to portray oneself in various social media platforms.

Partial Text

Visual experience is one of change, wherein people and objects constantly move about their environments. Motion signals are so inextricably linked to our perception of the world, that the brain has evolved neural modules exclusively dedicated to processing motion information [1–3]. Given the ubiquity of motion in our visual experience, it is no surprise that the role of motion in object perception has been extensively explored (e.g., [4, 5]). In face perception, for example, there is general consensus that motion confers a benefit to the recognition of familiar faces, a phenomenon referred to as the ‘motion advantage’ (e.g.,[6–8]). It most prominently manifests itself when viewing degraded images (e.g., negatives; [6]), ostensibly because under such viewing conditions recognition systems cannot rely upon cues typically available in a static image [9]. The visual system depends upon ‘dynamic facial signatures,’ cues present in moving faces, to identify faces in suboptimal viewing conditions [10]. This benefit is not unique to the identification of faces—emotional expressions are perceived as more salient in faces that are moving [11], further supporting a crucial role for motion in evaluating facial properties. These robust motion detection systems appear to be in place at an early age, as even infants show differential eye-movement patterns to faces in motion versus their static counterparts [12].

Experiment 1 explored whether the original FFE extended to other object categories, in this case, bodies.

Post and colleagues[18] found that the Frozen Face Effect persisted even when observers viewed inverted faces, suggesting that the FFE did not hinge upon holistic or configural information about the faces. In this experiment, we tested whether a similar pattern emerged with regard to bodies in motion. Given the relatively robust configural dependency found for bodies (on par with that of faces) [29], we might expect inverted bodies to reduce the magnitude of the Frozen Effect. However, we found an equivalent effect for both upright and inverted stimuli, similar to the original FFE, as described below.

It has previously been found that facial recognition is hindered when natural facial movement is disrupted [7, 13]. Similarly, disrupting coherent motion also reduces the Frozen Face Effect [18], suggesting that the benefit does not depend simply on a rapid succession of images or motion more generally. Experiment 1C explored whether coherent motion is driving the Frozen Effect observed in the first two experiments. To test this, observers evaluated the same upright images and movies from Experiment 1A, with the coherence of the depicted motion disrupted by scrambling the order of the frames in each movie. This isolated the role of coherent motion while controlling for other properties such as image speed, content, and number of images seen in a given unit of time.

Experiment 2 explored whether the predictability of motion modulated the Frozen Effect. We hypothesized that bodies moving in a prototypical fashion (e.g., a dancer executing a pirouette), when a given movement conforms to a constrained and predictable space, might elicit an even stronger Frozen Effect than that observed in the first experiment.

Given that flattery ratings were enhanced for both faces [18] and bodies in motion, we tested whether this effect generalized to other object categories as well, specifically, to non-biological objects often viewed in motion. We created a set of videos depicting moving objects, for example, machines, modes of transportation, and toys. The motion depicted in this stimulus set is even more constrained and predictable than the prototypical bodies-in-motion stimuli from Experiment 2, as bodies have far greater degrees of freedom than objects with well-defined motion patterns (e.g., vehicles, machinery, and automated toys). This implementation allowed us to assess whether the Frozen Effect generalized beyond biological entities.

In all previous experiments, the number of static images rated by participants was substantially greater than the number of videos rated (ratio of 30:1). In order to eliminate the possibility that participants were basing their flattering ratings on seeing a more novel stimulus (i.e., the video), the current experiment equated the overall number of static images and videos viewed by observers.

This experiment was a near complete replication of Experiment 4A, except that we equated the overall duration of the stimuli displayed on the screen. In Experiment 1, observers were exposed to the final frame of the vidoes for an extended period of time (until response), while in Experiments 2–4, the videos disappeared immediately upon completion while images remained on the screen until response. In this version, videos disappeared upon completion and images remained on the screen for one second before disappearing, which equated the overall amount of viewing time for each stimulus.

These experiments suggest that the ‘Frozen Effect’ extends well beyond the category of faces, and applies to a variety of both animate and inanimate objects in motion. The extent of this effect correlates with the predictability of depicted motion. Observers rated the motion stimuli as more flattering than the identical information presented in static form, whether it was children-at-play, bodies in prototypical motion, or objects that one typically associates with movement. The ubiquity of this effect reveals the importance of dynamic motion when making aesthetic judgments. While this effect was first revealed using faces[18], clearly motion plays a critical role in perceptual preference across a range of visual domains. The systematic increase in the Frozen Effect as a function of predictability, the design of Experiment 4, and the overall brevity of experimental sessions (around 30 minutes) rule out the possibility that these results reflect participant boredom.

Individuals rated bodies and objects in motion as more flattering than the identical information presented statically, something we refer to as the ‘Frozen Effect.’ This effect remained even when judging inverted stimuli, suggesting that an upright configuration of bodies and objects is not critical for eliciting the effect, consistent with the original Frozen Face Effect. When the motion of the stimulus was more predictable, the magnitude of the Frozen Effect increased. This may represent a kind of fault tolerance whereby bodies or objects with fewer constraints on the nature and direction of motion (e.g., children at play) do not appear aberrant. Bodies and objects that move in a prototypical fashion are more predictable, and therefore equivalent static depictions may be less aesthetically appealing.

All participants gave informed consent. This research, and all research described herein, was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board at Rhodes College.




Leave a Reply

Your email address will not be published.