The Role of Non-Representational Hand Gestures in Creative Thinking

Abstract

Previous studies suggest that representational gestures support divergent thinking and that mental imagery is necessary for gestures to aid convergent thinking. However, less is known about non-representational gesture use (i.e., beat and palm-revealing) during creative thinking. Across two experiments, we examined whether these gestures supported or hindered creativity and the effects of mental imagery on creative thinking. In Experiment 1, we tested both gesture-spontaneous and gesture-encouraged conditions during divergent thinking. Beat gestures, irrespective of condition, were negatively associated with originality in divergent thinking for individuals with high mental imagery. Encouraged palm-revealing gestures were negatively associated with fluency, flexibility, and elaboration in divergent thinking, regardless of mental imagery. In Experiment 2, we examined spontaneous gestures during both divergent and convergent thinking and assessed mental imagery vividness and skills. Beat gestures were negatively associated with convergent thinking for individuals with low or average imagery vividness. Similarly, palm-revealing gestures were negatively associated with convergent thinking for individuals with low mental imagery skills. Vividness of imagery was the only consistent positive predictor of divergent thinking. Spontaneous gestures were not associated with divergent thinking. These findings show that, unlike representational, non-representational gesture use does not facilitate and might even hurt creativity, depending on individual differences in mental imagery.

1. Introduction

Gestures are a fundamental component of human communication, closely linked to cognition and language. While extensive research has examined the cognitive benefits of representational gestures, such as iconic, metaphoric, and deictic gestures, which carry semantic information (e.g., Çapan et al., 2023; Clough & Duff, 2020; Güneş Acar et al., 2024; Hyusein & Göksun, 2023, 2024; Hostetter & Alibali, 2007, 2019; Kita et al., 2017; Novack & Goldin-Meadow, 2017), less is known about non-representational gestures, including beat and palm-revealing gestures (Chu et al., 2014; Müller, 2004, 2014). These gestures do not depict semantic content but instead serve mainly prosodic or pragmatic functions (McClave, 1994; Payrató & Teßendorf, 2014), raising questions about their potential contributions to cognitive processes. Representational gestures are known to support memory, problem-solving, and creativity by grounding thought in sensorimotor experience and interacting with one’s mental imagery (e.g., Aydin et al., 2023; Cook et al., 2011; Hyusein & Göksun, 2023, 2025). This paper investigates the self-oriented cognitive functions of non-representational gestures and asks whether they influence creative thinking, specifically divergent and convergent thinking, and whether these effects depend on individual differences in mental imagery skills and vividness. By examining spontaneous and encouraged gesture use across two experiments, we aim to clarify the cognitive relevance of non-representational gestures that are typically overlooked in the literature.

2. Gestures’ Role in Cognition

Much of the research on gestures and cognition has focused on representational gestures. Representational gestures convey semantic information about objects, actions, or spatial relations. Representational gestures include iconic, metaphoric, and pointing/deictic gestures (McNeill, 1992). Iconic gestures are used to depict the shape, movement, or features of real objects, actions, or events mentioned in speech, e.g., mimicking turning a steering wheel with the hands while saying “She swerved to the left to avoid the holes on the road.” Metaphoric gestures are used to depict abstract information, e.g., moving one hand up to “show” progress while saying “They improved a lot!” Last, pointing gestures, also called deictic gestures, include pointing with the index finger or with the whole hand to an object or an abstract reference (e.g., pointing at one’s temple when saying “She’s smart”), to draw attention to it. Gesture frameworks underlying the roles of gesture in cognition mainly focus on representational gestures. For example, the gesture-for-conceptualization hypothesis states that representational gestures facilitate thinking and speaking by activating, manipulating, packaging, and exploring spatio-motoric information (Kita et al., 2017). The Gesture as Simulated Action (GSA) framework also highlights that representational gestures arise from image-based motor simulation (Hostetter & Alibali, 2007, 2019). The account proposes that gestures occur when visual and motor areas of the brain activate to simulate mental images, which result in representational gestures, such as iconic and metaphoric. Empirical research provides support for these accounts. Representational gestures facilitate lexical fluency and verbal retrieval (e.g., Cook et al., 2011; Novack & Goldin-Meadow, 2017), memory (e.g., Aydin et al., 2023; Chu & Kita, 2011; So et al., 2011), and problem-solving (e.g., Hyusein & Göksun, 2023; Thomas & Lleras, 2009), among many cognitive functions.

Considerably less attention has been paid to non-representational gestures. Non-representational gestures do not depict semantic meaning but have pragmatic and interactive functions (Payrató & Teßendorf, 2014). Some studies show that they can also have a self-oriented function for the speaker’s thinking and speaking (e.g., Aydin et al., 2023; Hyusein & Göksun, 2023; Lucero et al., 2014). In this paper, we will focus specifically on beat and palm-revealing gestures. Beat gestures are rhythmic hand movements that align with speech prosody and are used to highlight or structure the speech discourse (McNeill, 1992; McClave, 1994). Palm-revealing gestures, also called palm-up-open-hand (PUOH) or simply palm-up gestures, are pragmatic gestures, characterized by an open hand with the palm facing upward (Chu et al., 2014; Müller, 2004, 2014). These gestures have been classified as palm-up presentational and palm-up epistemic by Cooperrider et al. (2018). The palm-up presentational gesture, comparable to Kendon’s (2004) “palm presentation”, is used to “present” something (a point) to the listener. The palm-up epistemic is what Chu et al. (2014) call a “palm-revealing” gesture and is used to convey absence of knowledge, uncertainty, and that the speaker has nothing more to say. Moreover, Müller (2004) distinguishes another function of PUOH gestures—expressing obviousness, e.g., when the speaker is saying “How could it be otherwise?” or “What else could one say?”. Here, we will focus on palm-up epistemic gestures that are used to express absence of knowledge, uncertainty, and obviousness and will use the name “palm-revealing” gestures for simplicity.

2.1. The Role and Cognitive Functions of Beat and Palm-Revealing Gestures

There has been a debate over the functions and overlap of non-representational gestures with representational ones. For example, researchers have argued that the origin of palm-revealing gestures is based on a metaphorical action of giving and receiving objects or ideas (Müller, 2004, 2014; McNeill, 1992). Beat gestures have also been suggested to be simplified forms of iconic gestures (Feyereisen & Havard, 1999). This argument is based on the developmental course of gestures, with representational gestures appearing earlier in children compared to non-representational ones. The first beat gestures appear around the age of five and increase in adolescence (Freedman et al., 1986). This could be due to the later development of abstract thinking and the use of complex syntax structures (Feyereisen & Havard, 1999). More recent empirical work supported the notion that beat gestures might be lexically integrated later in language development (e.g., Prieto et al., 2018; Rohrer et al., 2020; So et al., 2011; Vilà-Giménez & Prieto, 2020). For example, So et al. (2011) found that beat gestures aid memory recall for adults but not yet for children. Even if beat gestures were remnants of representational gestures, they could simply be used to keep the rhythm of the prosody and aid complex sentence structures by cognitively offloading them to the hands or focusing attention on critical parts of the speech, as suggested by more recent literature reports (Biau et al., 2018; Holle et al., 2012; Türk & Calhoun, 2023). McNeill (1992) argued that even if beat gestures do not have a visual or semantic content, they might be superimposed with iconic gestures, thus simultaneously taking on an iconic meaning. Empirical evidence by Yap et al. (2018) supported the idea that beats and representational gestures might emerge from the same system and, in some cases, might be superimposed. In their study, participants’ beat gestures moved in the direction that the spatial schemas of the story appeared (e.g., grades improving or a rocket taking off).

Moreover, in contrast to representational gestures, non-representational gestures are believed to be non-imagistic or unrelated to mental imagery. For example, according to McNeill (1992), if gestures are placed on a continuum with varying degrees of iconicity, beat gestures would fall on the low iconicity end because they have lost their imaged content. Krauss et al. (1995) argued that there were two separate components underlying the production of representational and non-representational gestures. While representational gestures depend on the spatio-dynamic motoric functions of working memory and rely on visual imagery, non-representational gestures are connected to the phonological encoding of sentences (Bosker & Peeters, 2021) and do not rely on visual mental imagery (Arslan & Göksun, 2020).

Research on the speaker-oriented function of beat gestures has shown that using beats can benefit thinking. For example, encouraging children to produce beat gestures in the training phase of a narrative task improved both their fluency and the narrative structure (Vilà-Giménez & Prieto, 2018). A different study showed that instructing people to produce beat gestures when guessing a word from its meaning boosted their response times compared to producing iconic gestures (Lucero et al., 2014). Moreover, even if metaphoric gestures were used to recall space-, motion-, and object-based metaphors, it was beat gestures that were used when recalling static event-based metaphors (Khatin-Zadeh et al., 2022). Beat gestures have also been associated with remembering external details for past and future events in autobiographical memory (Aydin et al., 2023). The latter might relate to creativity as well, because if beat gestures contribute to the embellishment of memories, then they might also contribute to the elaboration of ideas, which is an important component of creative thinking.

Palm-revealing gestures’ effect on thinking has not been studied so extensively because they are mostly considered to be pragmatic or interactive gestures used for the listener and are less associated with a speaker-oriented function. For example, Pang et al. (2024) showed that individuals were more likely to perceive replies as less straightforward and more indirect when the replies were accompanied by palm-revealing gestures or other nonverbal behaviors signaling uncertainty, e.g., head tilt, gaze aversion, or a face shrug. Palm-revealing gestures are also known to have an epistemic function—when people reach an “impasse”, palm-revealing gestures are used to reassert a position as being obvious (Marrese et al., 2021). More recent research showed that the use of palm-revealing and beat gestures (combined into a non-representational gesture category) when people experienced disfluency of speech impacted metacognitive judgments (Yılmaz et al., 2025). In particular, individuals who were disfluent during a general knowledge task and used more non-representational gestures had higher retrospective confidence judgments (i.e., self-rated confidence on the general knowledge answers). This shows that non-representational gestures might falsely boost confidence, but more research needs to be performed to understand if palm-revealing gestures have any beneficial effects on cognitive functions, both for speakers and listeners.

An important theoretical question, which the current literature has not yet fully elucidated, concerns the directionality of the gesture–cognition relationship. On the one hand, beat and palm-revealing gestures might shape the underlying thought processes. For example, beat gestures can scaffold attention (e.g., Dimitrova et al., 2016; Ongchoco & Scholl, 2021), or palm-revealing gestures can offload cognitive load. Alternatively, gestures might be indicators of the speaker being engaged in a thought process, and, similar to disfluencies, gestures might be used to signal to the listener that the speaker is thinking and will soon produce a message/answer (Clark & Fox Tree, 2002; Yılmaz et al., 2025). These perspectives are not mutually exclusive, and the same gesture may serve both roles at the same time. For example, a beat gesture might be used to maintain the flow of the conversation and engage the listener’s attention while helping the speaker focus their attention on the task. In other words, gestures can both reflect and alter thinking (Goldin-Meadow & Beilock, 2010), and a directional cause-and-effect relationship might be difficult to infer.

Based on previous theories and findings regarding non-representational gestures, in this paper, we ask whether non-representational gestures have a similar function for higher-order cognition, such as creative thinking, in a way that representational gestures do. Moreover, we will also try to unpack the role of mental imagery for non-representational gestures during creative problem-solving tasks.

2.2. The Role of Gestures in Creative Thinking

In this paper, we focus on a less-studied but equally important cognitive function that might be related to and enhanced by gestures—creativity. Creativity is generally defined as the ability to generate novel and useful ideas or products (Runco & Jaeger, 2012). It has recently been studied under the embodied cognition account, and a number of studies have shown that movement can benefit creative output. For example, Rominger et al. (2020) found that everyday bodily movement was associated with creative performance. Hand movements, such as fluid arm movement and squeezing a ball with the hands, were also associated with enhanced creativity (e.g., Kim, 2015; Slepian & Ambady, 2012). Even more specifically, representational hand gestures were shown to benefit children’s fluency (Kirk & Lewis, 2017) and narration skills (Laurent et al., 2020) and young adults’ fluency, originality, and elaboration in divergent thinking, defined as the ability to come up with many different and original solutions to a problem (Hyusein & Göksun, 2024). Interestingly, for convergent thinking, which is the ability to find the best solution to a problem, mental imagery interacted with gestures in a way that people with higher imagery skills benefited from gestures, while those with low imagery did not and their convergent thinking was even negatively affected by their gestures (Hyusein & Göksun, 2023). These findings were especially true for representational gestures. However, Hyusein and Göksun (2023) also found that beat gestures could benefit creativity if, in addition to high imagery skills, people also had more practice with the convergent thinking task. Similar to representational gestures, the use of encouraged beat gestures hampered creativity in people with low imagery. This is an intriguing finding because even though it was proposed that beat gestures, in comparison to representational gestures, have lost their imagistic content (McNeill, 1992), they still rely on individuals’ imagery. Similarly, palm-revealing gestures were also negatively associated with convergent thinking scores in individuals with low and average imagery ability. It is possible that the lack of knowledge or uncertainty was reflected in those gestures. However, it is not clear whether low imagers used more palm-revealing gestures compared to high imagers. In any case, palm-revealing gestures did not affect high imagers’ convergent thinking as beat gestures did; therefore, beat gestures might be more crucial for the speaker’s thinking and problem-solving, and palm-revealing gestures might mostly be used for the listener. Another study that found an association between creativity and beat gestures focused on improvisational skills (Lewis et al., 2015). Even though improvising elicited a higher number of iconic and deictic gestures compared to everyday speech, it was beat gestures, in addition to iconic gestures, that were associated with a higher quality of improvisation.

Despite research showing that beat gestures might play an important role in creativity, non-representational co-speech gestures have been mostly excluded in previous studies and were even considered non-creative (Cienki & Mittelberg, 2013; Marschark et al., 1986). For example, Marschark et al. (1986) studied linguistic creativity in deaf and hearing children while excluding pointing and beat gestures.

2.3. The Role of Mental Imagery in Gestures and Creative Thinking

Mental imagery is a sensorimotor experience that occurs in the absence of external stimuli (Donoff, 2019). It enables us to interact and manipulate internal simulations and thus perform a wide range of tasks, such as spatial navigation (Palermo et al., 2008), mental rotations (Shepard & Metzler, 1971), or future planning (Moulton & Kosslyn, 2009). Mental imagery is also pivotal for creativity and imagination (e.g., the Geneplore Model of Creativity, Finke et al., 1992; Salvi & Bowden, 2016). Gesture frameworks have already pointed out its relevance for gestures, with representational gestures being at the forefront (e.g., the gesture-for-conceptualization hypothesis, Kita et al., 2017). Research has emphasized the multi-faceted nature of mental imagery and the need to distinguish between mental imagery skills related to visuospatial manipulations and vividness of mental imagery, which could mark how closely related the imagined object is to the real percept or how immersed in the experience the individual is (Hyusein & Göksun, 2025; McAvinue & Robertson, 2007). The vividness of mental imagery was also associated with convergent thinking in such a way that a more vivid mental imagery during the creativity task, along with representational gesture use, indicated lower convergent thinking performance (Hyusein & Göksun, 2025). This finding contradicts Hyusein and Göksun (2023), who found a positive association between representational gestures and convergent thinking for people with high mental imagery skills. This is an example of how mental imagery skills might be different from mental imagery vividness.

2.4. The Present Study

The present research investigates the role of non-representational gestures—specifically, beat gestures and palm-revealing gestures—in creative thinking. Across two experiments, we examine both spontaneous and encouraged gesture production and assess their associations with performance in divergent thinking (generating many ideas) and convergent thinking (finding a single correct solution). We also explore how these relationships are moderated by individual differences in mental imagery skills and the vividness of mental imagery.

In Experiment 1, participants generated creative uses for everyday objects under spontaneous and encouraged gesture conditions, and we assessed the predictive role of gesture production alongside mental imagery abilities for divergent thinking only. In Experiment 2, focusing on spontaneous gestures only, we measured both divergent and convergent thinking and examined the interplay between gesture use, imagery skills, and vividness.

3. Experiment 1

In this experiment, we used the dataset of Hyusein and Göksun (2024) to analyze the role of spontaneous and encouraged non-representational gestures in divergent thinking. The study by Hyusein and Göksun (2024) focuses on iconic gestures only. Here, we examined the beat and palm-revealing gestures of the same participants, accounting for individual differences in mental imagery. We hypothesize the following:

1.

Encouraged beat gestures will be helpful for divergent thinking, irrespective of mental imagery skills, as beat gestures have been proposed to be less imagistic, and divergent thinking was not related to mental imagery in previous studies (see Hyusein & Göksun, 2024).

2.

Both spontaneous and encouraged palm-revealing gestures will be negatively associated with divergent thinking because they serve interactive functions.

3.1. Method

3.1.1. Participants

The sample consisted of 80 young adults (M_age = 21.3, SD = 2.46; 52 females) randomly assigned to either a gesture-spontaneous + gesture-encouraged condition (Group 1; N = 40) or a gesture-encouraged condition only (Group 2; N = 40). The number of participants was based on earlier research examining how gestures influence creative thinking (e.g., Kirk & Lewis, 2017; Liao & Wang, 2019). For instance, Kirk and Lewis (2017) determined their sample size based on a power analysis targeting 80% power to detect a medium effect size (dz = 0.5). The study received ethical clearance from the Koç University Committee on Human Research (IRB Protocol No. 2020.125.IRB3.063).

3.1.2. Materials

Divergent Thinking Assessment

Alternative Uses Test (AUT; adapted from Guilford, 1967)

To assess divergent thinking, participants were asked to list as many creative uses as possible for common household items within a 1-min time limit per item. Four objects were used in total—newspaper, pencil, shoe, and towel—with two assigned to each condition in a counterbalanced order.

Participants’ responses were evaluated along four dimensions:

Fluency—Total number of responses. Each idea was given 1 point.

Originality—Ideas were evaluated based on their rarity across the sample. Responses given by 5% of the participants were labeled as “unusual” (1 point), and those given by just 1% were considered “unique” (2 points). The rest of the responses that were elicited by more than 5% of the sample were given 0 points.

Flexibility—The count of distinct semantic categories represented in the responses. Each new category earned 1 point; if a response repeated a previously used category, it received 0 points.

Elaboration—The degree of descriptive detail. For instance, the response “a boat” scored 0, while “I could use a shoe as a boat, by putting toys in it, in a pool” scored 2 points—one for the functional explanation and another for added context.

Mental Imagery Assessment

Mental Imagery Test (MIT; Di Nuovo et al., 2014)

Mental imagery skills were measured using an 8-task battery assessing image generation, retention, and transformation. Tasks required participants to mentally manipulate visual information. For example, in one task, participants were shown a 3 × 3 cube structure (9 mini-cubes per face) with visible external faces that were colored pink. After 30 s of viewing, the picture of the cube was removed from participants’ sight, and they had to recall how many cubes had three, two, one, or no colored sides without looking at the picture. The MIT was previously used in Turkish samples using translated instructions (Arslan & Göksun, 2020; Hyusein & Göksun, 2023). The tasks were displayed on a white background, and responses were recorded manually by the experimenter. A total MIT score was calculated by summing all task scores, with a maximum possible score of 83.

3.1.3. Procedure and Study Design

Due to the global COVID-19 pandemic, which coincided with this study’s data collection process, all data were collected online through Zoom by the first author of the paper and three trained research assistants. All testing sessions were recorded to enable later analysis of speech and gesture. After obtaining verbal consent, participants were asked to position themselves so that their upper body, arms, and hands were fully visible on the camera.

Group 1 completed the AUT twice—once in a gesture-spontaneous condition, where no instructions were given about hand use, and once in a gesture-encouraged condition, where they were specifically asked to use their hands during the task. Group 2 completed the AUT only in the gesture-encouraged condition. This design allowed both within-subject (Group 1) and between-subject (comparing Group 1 and Group 2) analyses, helping to distinguish effects of gesture encouragement from potential learning effects. The prompt for the gesture-encouraged condition was: “Please, use your hands while speaking and thinking when solving the next task.”

After completing the creativity and mental imagery tasks, participants completed a demographic form, were debriefed about the purpose of the study, and were granted a course credit to thank them for their time.

3.1.4. Coding and Scoring Procedures

Alternative Uses Test (AUT) Scoring

Responses from the AUT were transcribed into Microsoft Excel by the first author and a trained research assistant. For details regarding the transcriptions, see the Coding and Scoring section of Hyusein and Göksun (2024). To assess coding consistency, interrater reliability was calculated for fluency, flexibility, and elaboration on 20% of the dataset, independently coded by both the first author and a research assistant. The intra-class correlation coefficients (ICCs) indicated excellent agreement across all three measures—fluency: ICC = 0.971, 95% CI [0.936, 0.989], p < 0.001; flexibility: ICC = 0.970, 95% CI [0.934, 0.985], p < 0.001; elaboration: ICC = 0.968, 95% CI [0.942, 0.984], p < 0.001. Since originality scores were calculated via a standardized procedure, interrater reliability for that metric was not necessary.

Gesture Coding

As participants’ speech was already transcribed in Microsoft Excel, gestures were also coded in the same files by a trained research assistant. Gestures were coded into 6 separate categories—iconic, deictic, metaphoric, beat, palm-revealing, and other. The latter included emblem gestures and gestures that did not fit into the other five categories. Each gesture was assigned exclusively to one category (i.e., categories were mutually exclusive). To ensure consistency and reliability in gesture coding, a second assistant cross-checked the gesture annotations, and any discrepancies were reviewed and resolved by the first author. Then, a third independent assistant coded the gestures of 20% of the dataset. The agreement on the number of occurrences of all gesture types between the coders was excellent (ICC = 0.910, 95% CI [0.869, 0.939], p < 0.001).

Gesture frequency was calculated by dividing the number of gestures by the total number of words spoken for each trial. This normalized rate (rather than raw counts) was used to account for individual differences in speech production, as longer speech naturally leads to more gestures (So et al., 2009).

All data can be accessed via the Open Science Framework: https://osf.io/c7umf/?view_only=3e2eef623843477f8ccb86dcdcde02cb (accessed on 12 August 2025).

3.2. Results

Descriptive statistics (mean and standard deviation) for the divergent thinking measures (fluency, originality, flexibility, and elaboration), as well as gesture types (i.e., beat and palm-revealing gestures), are presented in

Table 1. Mean (M) and standard deviation (SD) values of AUT/divergent thinking measures, beat and palm-revealing gestures, and mental imagery scores across groups and conditions.

	Group
	Gesture-Spontaneous Group 1 (GS1)		Gesture-Encouraged Group 1 (GE1)		Gesture-Encouraged Group 2 (GE2)
	M	SD	M	SD	M	SD
AUT Fluency	6.09	2.12	5.90	1.81	5.63	2.29
AUT Originality	2.48	1.97	2.49	1.94	2.44	2.33
AUT Flexibility	4.31	1.56	4.06	1.24	3.63	1.53
AUT Elaboration	5.94	3.14	7.09	3.77	5.26	3.13
Beat Gesture Frequency	0.04	0.05	0.05	0.05	0.05	0.06
Palm-revealing Gesture Frequency	0.01	0.02	0.02	0.03	0.02	0.05
Mental Imagery Score	69.42	7.32	69.42	7.32	68.83	7.58

.

To analyze the data, we used the R statistical software version R 4.3.2 (R Core Team, 2013). All continuous variables were standardized before performing the analysis with the scale() function. To test within-subject effects for Group 1, we performed repeated-measure ANCOVAs with the aov() function. Participant ID was included as a random factor to account for the within-subject variance across conditions. To test between-effects for Group 1 and Group 2, we also used the aov() function but without random factors. The summary() function was used to obtain F-statistics and p-values. Partial eta squared values to estimate effect sizes were obtained with the eta_squared() function. To test interaction effects involving continuous variables, we used the emtends() function from the emmeans package. To visualize interaction and main effects, we used the interact_plot() and ggplot() functions.

3.2.1. Differences in Non-Representational Gesture Frequency Across Groups and Conditions

Group 1 produced 138 beat and 55 palm-revealing gestures in their gesture-spontaneous and 193 beat and 77 palm-revealing gestures in their gesture-encouraged condition. Group 2, who attended a gesture-encouraged condition only, produced 248 beat and 65 palm-revealing gestures. The differences in beat and palm-revealing frequencies between the groups and conditions were not statistically significant (see Tables S1 and S2 in the Supplementary Materials).

3.2.2. Beat Gestures in Group 1 (Gesture-Spontaneous and Gesture-Encouraged Conditions)

First, we tested the hypothesis that encouraged beat gestures would be beneficial for divergent thinking, irrespective of mental imagery skills. We built separate ANCOVA models with the four divergent thinking dimensions (fluency, flexibility, originality, and elaboration) as dependent variables and condition, beat gestures, mental imagery, and the two-way and three-way interactions between them as predictor variables. To account for within-subject effects, we included participant ID as a random factor across conditions.

The model predicting originality scores included a significant two-way interaction between beat gestures and mental imagery, F(1,61) = 6.51, p = 0.013, η² = 0.10. Simple slope analysis indicated that only the slope for high mental imagery (+1 SD) was significant, b = −0.31, SE = 0.13, t(129) = −2.51, p = 0.014 (see Figure 1). Irrespective of the gesture condition, for people with higher mental imagery, beat gestures were associated with decreased originality.

There was also a main effect of condition (gesture-spontaneous vs. gesture-encouraged) on elaboration, F(1,28) = 6.98, p = 0.013, η² = 0.20. Estimated marginal means indicated that elaboration was significantly higher in the gesture-encouraged condition (M = 7.24, SE = 0.53) compared to the gesture-spontaneous condition (M = 6.17, SE = 0.53), regardless of gesture frequency and mental imagery skills.

Repeated-measures ANCOVAs for fluency and flexibility scores did not yield any significant effects.

Overall, beat gestures were not beneficial for divergent thinking, and they were even associated with reduced originality of ideas when produced by individuals who had high mental imagery.

3.2.3. Beat Gestures in Group 1 and Group 2 (Gesture-Encouraged Conditions)

Next, we tested the same hypothesis (H1: Encouraged beat gestures will be beneficial for divergent thinking, irrespective of mental imagery skills) with between-subject effects of Group 1 and Group 2’s gesture-encouraged conditions.

Similar to the within-subject analysis, we found an interaction effect between beat gestures and mental imagery irrespective of group, F(1,129) = 5.55, p = 0.02, η² = 0.04. Simple slope analysis revealed that only the slope for high mental imagery (+1 SD) individuals was significant, b = −0.33, SE = 0.13, t(129) = −2.51, p = 0.013, irrespective of the gesture condition. Individuals with high mental imagery performed worse on the RAT when using beat gestures.

As in the within-subject analysis of Group 1, between-subject effects indicated a main effect of group (Group 1 vs. Group 2) on creative elaboration, F(1,129) = 10.63, p = 0.001, η² = 0.08. Pairwise comparisons of the raw elaboration scores with the Bonferroni adjustment showed that Group 1 (M = 7.19, SE = 0.42) produced significantly more elaborate responses compared to Group 2 (M = 5.39, SE = 0.41).

As in the within-subjects analysis, between-subject ANCOVAs did not yield any significant effects of beat gestures or mental imagery on fluency and flexibility when gestures were encouraged.

Both the within- and between-group analyses show that beat gestures’ function for originality of ideas does not change with gesture manipulation, and an increased use of beats reflects a lack of originality.

3.2.4. Palm-Revealing Gestures in Group 1 (Gesture-Spontaneous and Gesture-Encouraged Conditions)

Repeated-measures ANCOVA with fluency as the dependent variable and condition, palm-revealing gestures, mental imagery, and their interactions as predictors revealed a significant two-way interaction between condition and palm-revealing gestures, F(1,28) = 5.06, p = 0.03, η² = 0.15. However, simple contrasts did not yield a significant difference between the two conditions, b = 0.48, SE = 0.25, t(125) = 1.916, p = 0.058. The simple slopes of the gesture-encouraged condition (b = 0.28, SE = 0.18, t(126) = 1.606, p = 0.111) and the gesture-spontaneous condition (b = −0.20, SE = 0.18, t(125) = −1.105, p = 0.271) were also not significant (see Figure 2).

There were no significant effects of condition, palm-revealing gestures, or mental imagery on flexibility, originality, or elaboration.

3.2.5. Palm-Revealing Gestures in Group 1 and Group 2 (Gesture-Encouraged Conditions)

The between-subject ANCOVA with flexibility as the dependent variable and group, palm-revealing gestures, mental imagery, and their interactions as the predictor variables indicated a significant two-way interaction between group and palm-revealing gestures, F(1,129) = 8.79, p = 0.004, η² = 0.06. Simple slopes analysis revealed that the slope of Group 2 was significant (b = −0.49, SE = 0.14, t(129) = −3.37, p = 0.001), indicating that there was a negative association between the palm-revealing gestures of people in Group 2 and their flexibility (see Figure 3). The slope of Group 1 did not reach significance, b = 0.17, SE = 0.17, t(129) = 1.00, p = 0.319.

Next, the between-subject ANCOVA with elaboration as the dependent variable and group, palm-revealing gestures, mental imagery and their interaction as the predictor variables revealed significant main effects of group, F(1,129) = 10.68, p = 0.001, η² = 0.03, and palm-revealing gestures, F(1,129) = 4.84, p = 0.03, η² = 0.03. As previously indicated by the between-subject analysis for beat gestures, individuals in Group 1 (M = 7.19, SE = 0.42) produced significantly more elaborate responses compared to those in Group 2 (M = 5.39, SE = 0.41). The main effect of palm-revealing gestures indicated that there was a negative association between palm-revealing gesture use and elaboration scores for both groups (see Figure 4).

There were no effects of palm-revealing gestures or mental imagery for fluency or originality.

3.3. Discussion

The present experiment investigated whether non-representational gestures, specifically beat and palm-revealing, supported divergent thinking and whether these effects depended on individual differences in mental imagery skills. Contrary to our initial hypotheses, non-representational gestures were not positively associated with divergent thinking and, under certain conditions, they might have reflected cognitive strain or reduced creative output.

First, we hypothesized that encouraged beat gestures would benefit divergent thinking regardless of mental imagery ability due to their reduced imagistic properties. This was not supported. Instead, beat gestures were associated with significantly lower originality scores in individuals with high mental imagery across both within- and between-subject analyses. We also found a consistent main effect of gesture condition on elaboration. Participants produced more elaborate responses in the gesture-encouraged condition than in the spontaneous one.

Regarding palm-revealing gestures, our second hypothesis predicted that they would be negatively associated with divergent thinking because of their interactive and non-self-oriented role. The results partially supported this. While the within-subject analysis suggested a significant interaction between condition and palm-revealing gestures with a large effect size, suggesting that there was a robust divergence in the effects of gestures on fluency depending on the gesture condition, the individual slopes of the conditions were not statistically significant. The between-group comparison revealed a significant negative association between palm-revealing gesture frequency and flexibility in Group 2. Furthermore, there was a significant main effect of palm-revealing gestures on elaboration, showing a negative association across both groups’ gesture-encouraged conditions. These findings suggest that when individuals are prompted to gesture but experience difficulty generating creative responses, palm-revealing gestures may emerge as a pragmatic cue reflecting uncertainty or lack of knowledge.

To further investigate the role of non-representational gestures in creative thinking, in Experiment 2, we examined non-representational gestures in both divergent and convergent thinking and vividness of mental imagery during creative thinking.

4. Experiment 2

In the second experiment, we used the dataset of Hyusein and Göksun (2025) to analyze the role of spontaneous non-representational gestures in divergent and convergent thinking while accounting for individual cognitive differences in mental imagery skills and vividness. The study of Hyusein and Göksun (2025) focuses on representational gestures elicited during two conditions—Turkish and English. Here, we will examine the beat and palm-revealing gestures of the same participants during the Turkish condition only to compare the findings better with Experiment 1.

We hypothesized the following:

• Spontaneous beat and palm-revealing gestures will be negatively associated with divergent thinking, as shown in Experiment 1.
• Spontaneous beat and palm-revealing gestures will not be associated with convergent thinking (based on Hyusein and Göksun’s (2023) findings).
• Higher mental imagery vividness might negatively impact the relationship between spontaneous beat and palm-revealing gestures and convergent thinking (based on Hyusein & Göksun, 2025).

4.1. Method

4.1.1. Participants

The sample size of Hyusein and Göksun’s (2025) study, N = 40, was determined by power analysis (the alpha level was set as 0.05, the effect size as (np2) 0.25, and the power as 0.80). Here, we analyzed the data of 38 participants (M_age = 21.75, SD = 2.43, 19 female) who attended the Turkish condition. Participants were undergraduate and graduate students from Koç University and received either course credit or monetary compensation for their time. The study was granted ethical approval by Koç University’s Institutional Review Board committee with the approval number 2023.022.IRB3.006.

4.1.2. Materials

As in Experiment 1, participants completed the AUT (Guilford, 1967) to measure divergent thinking and the MIT (Di Nuovo et al., 2014) to assess mental imagery skills. In this study, the time limit per object for the AUT was 3 min. To measure convergent thinking, we administered Mednick’s Remote Associates Test (RAT; Mednick, 1962). Participants were presented with 10 groups of three words and were instructed to find a fourth common word for each triad in 30 s each. For example, for the triad “cottage, Swiss, cake”, the common associate is “cheese”. The words were presented on a white screen and were read out loud by the experimenter. Scoring was based on the number of correct responses—each correct response scored as 1.

To assess the vividness of mental imagery during the AUT and the RAT, we asked participants to self-report their imagery vividness after each trial by answering the question, “While doing this task, you may have had some imagery experiences evoked. Please rate by advancing the line in the scale below, the clarity and distinctness with which you were able to generate images while solving this task”, on a scale of 0–100.

4.1.3. Coding Procedure

AUT. Responses were transcribed and coded by two trained research assistants. The interrater reliability for 20% of the data was calculated for fluency, flexibility, and elaboration. The interrater intra-class correlation coefficients (ICCs) showed almost perfect agreement for all three measures, ICC_fluency = 0.999, 95% CI [0.998, 1.000], p < 0.001; ICC_flexibility = 0.926, 95% CI [0.854, 0.963], p < 0.001; ICC_elaboration = 0.888, 95% CI [0.785, 0.944], p < 0.001. Originality scores were calculated based on a standardized procedure, similar to Experiment 1.

Gestures. Speech and gestures for the RAT were transcribed and coded on the ELAN language archive software (version 6.5; 2023) by a trained research assistant. As in Experiment 1, gestures were coded into six categories—iconic, deictic, metaphoric, beat, palm-revealing, and other (e.g., emblems or gestures that did not fit any of the categories above). Excerpts from the ELAN coding screen with example beat and palm-revealing gestures are depicted in Figure 5. Each gesture was assigned exclusively to one category (i.e., categories were mutually exclusive). The first author coded 20% of the data to ensure interrater reliability. There was excellent agreement between the two coders, ICC = 0.839, 95% CI [0.765, 0.891], p < 0.001.

Speech and gestures for the AUT were transcribed and coded in Microsoft Excel by two trained research assistants. The same types of gestures as in the RAT were coded. Any ambiguities pointed out by the assistants were resolved by the first author. Interrater reliability calculated for 20% of the data indicated an excellent agreement for categorizing gesture frequency across trials (ICC = 0.964, 95% CI [0.949, 0.982], p < 0.001) between the two coders.

For both tasks (RAT and AUT), the gesture frequency rate was calculated for each gesture type by dividing the number of gestures produced by the number of words spoken in each trial. In the current experiment, we only used the beat and palm-revealing gesture frequency rates, as those are the focus of the current study.

All data are available on the Open Science Framework: https://osf.io/4paf2/?view_only=9ad2d853d04946da943bcfd19af1a2c1 (accessed on 12 August 2025).

4.2. Results

The descriptive statistics (i.e., mean, standard deviation, minimum, and maximum) of the AUT, RAT, vividness of mental imagery, non-representational gesture frequency rates, and MIT scores can be found in

Table 2. Mean (M), standard deviation (SD), minimum (Min.), and maximum (Max.) values of the Alternative Uses Task (AUT), Remote Associates Test (RAT), vividness of mental imagery ratings during AUT and RAT, non-representational gesture frequency rates during AUT and RAT, and Mental Imagery Test Scores.

	M	SD	Min.	Max.
AUT Score	71.7	27.2	24	127
RAT Score	6.4	1.7	2	10
AUT Vividness	70.9	20.1	31	100
RAT Vividness	55.0	20.1	12.6	94.5
AUT Beat Gesture Frequency	0.05	0.04	0	0.18
RAT Beat Gesture Frequency	0.03	0.05	0	0.25
AUT Palm-revealing Gesture Frequency	0.01	0.01	0	0.05
RAT Palm-revealing Gesture Frequency	0.03	0.04	0	0.17
Mental Imagery Score	71.9	7.03	54	83

.

In Experiment 1, we employed ANOVA models to examine both within- and between-subject effects across two groups and two conditions. In contrast, Experiment 2 involved a single group of participants tested under a single condition, with continuous predictors (e.g., gesture frequency, mental imagery skills, and vividness). Therefore, we used linear regression to model the relationship between these continuous variables and the outcome measures, as it is more appropriate for probing interactions and estimating slopes in such designs.

As in Experiment 1, we analyzed the data in the R statistical software version R 4.3.2 (R Core Team, 2013). All continuous variables were standardized before running the analysis with the scale() function.

4.2.1. Beat Gestures During Divergent Thinking

We built separate models for each divergent thinking measure (fluency, flexibility, originality, elaboration) as a dependent variable. We included beat gestures during the AUT as the first predictor and either imagery vividness during the AUT or mental imagery skills as the second predictor, as well as their interaction effects. None of the models indicated significant main or interaction effects of spontaneous beat gestures or mental imagery on divergent thinking.

In Experiment 1, we observed a significant interaction between beat gestures (irrespective of the gesture condition) and mental imagery skills for originality in divergent thinking. To test whether the failure to replicate this finding in Experiment 2 was due to insufficient statistical power, we conducted post hoc power analysis using the pwr.f2.test() function from the pwr package in R. We found that the power of the current sample was lower than 0.80, power = 0.69. Therefore, the current study might have been slightly underpowered to detect an interaction between beat gestures and mental imagery. Nevertheless, visual inspection of the interaction plot between beat gestures and mental imagery did not yield any discernible pattern that would suggest a trend toward an interaction. Therefore, this may indicate a genuine absence of an effect in the context of the current experiment, which is not due to insufficient power.

4.2.2. Beat Gestures During Convergent Thinking

We built two separate models for convergent thinking as the dependent variable. In the first model, we included beat frequency during the RAT and mental imagery skills and their interaction as predictors. In the second model, we replaced mental imagery skills with mental imagery vividness during the RAT. The first model did not yield any significant effects. However, in the second model, both the main effects and the interaction term were significant. We further inspected the interaction between beat gestures and mental imagery vividness during the RAT, b = 0.40, SE = 0.16, t(34) = 2.55, p = 0.015, by estimating simple slopes. Simple slopes analysis revealed that beat gesture frequency was negatively associated with RAT performance at low (b = −0.59, p = 0.02) and mean levels of vividness (b = −0.21, p = 0.04), whereas higher vividness was not associated with better RAT performance when beat gestures were used (b = 0.16, p = 0.012) (see Figure 6).

4.2.3. Palm-Revealing Gestures During Divergent Thinking

Similar to the analysis for beat gestures during divergent thinking, we built models for the different dimensions of convergent thinking as dependent variables and palm-revealing gesture frequency during the AUT and mental imagery skills or vividness during the AUT as predictors, as well as their interactions. In all models, only the main effect of vividness of mental imagery during the AUT was a significant predictor of AUT performance. Specifically, higher vividness of mental imagery during the AUT significantly predicted better performance on elaboration (b = 0.72, SE = 0.16, t = 4.56, p < 0.001), originality (b = 0.63, SE = 0.17, t = 3.65, p < 0.001), flexibility (b = 0.63, SE = 0.18, t = 3.50, p = 0.001), and fluency (b = 0.65, SE = 0.16, t = 4.02, p < 0.001). No significant effects were found for palm-revealing gestures or their interactions with imagery vividness.

4.2.4. Palm-Revealing Gestures During Convergent Thinking

We built two separate models to test the effects of palm-revealing gestures on convergent thinking. The first model included an interaction between palm-revealing gestures and mental imagery skills, and the second one included an interaction between palm-revealing gestures and mental imagery vividness during the RAT. In the first model, we detected a significant interaction between gestures and vividness, b = 0.70, SE = 0.31, t(32) = 2.27, p = 0.03. Simple slopes analysis revealed that palm-revealing gesture frequency was negatively associated with RAT performance only at low levels of vividness, b = −0.97, p = 0.03 (see Figure 7). There were no significant effects in the second model (palm-revealing gestures x imagery vividness).

4.3. Discussion

This experiment investigated the relationship between spontaneous non-representational gesture use and divergent and convergent thinking, as moderated by mental imagery skills and task-specific mental imagery vividness.

None of the models in the present study yielded significant main or interaction effects for beat gestures or imagery skills, or vividness during the AUT for divergent thinking. For convergent thinking (RAT performance), we found a significant interaction between beat gesture frequency and vividness of mental imagery during the RAT. Beat gestures were associated with lower RAT scores at low and average levels of vividness but not at high levels.

For divergent thinking, palm-revealing gestures did not interact with mental imagery. However, vividness of mental imagery during the AUT emerged as a consistent positive predictor across all divergent thinking dimensions. This was not the case in the beat gesture models, raising questions about whether palm-revealing gestures interact less with cognitive processes and are merely used for the listener as interactive gestures. A post hoc correlational analysis showed that while beat gestures were moderately negatively correlated with vividness (r = −0.34, p = 0.04), palm-revealing gestures were not correlated with vividness (r = −0.14, p = 0.41). This suggests that in our beat gesture model, there was a multicollinearity between gestures and vividness, which suppressed the effect of imagery vividness on divergent thinking, i.e., beat gestures and vividness shared variance in the model.

Moreover, palm-revealing gestures during the RAT showed a significant interaction with mental imagery ability: they were negatively associated with RAT performance at low levels of imagery. This mirrors the interaction found for beat gestures and suggests that both types of non-representational gestures may signal or induce cognitive difficulty in individuals who lack vivid internal representations.

5. General Discussion

Across two experiments, we examined whether non-representational gestures—specifically beat and palm-revealing gestures—supported or hindered creative thinking and how these effects were shaped by mental imagery. We discuss our findings for beat and palm-revealing gestures separately.

5.1. Beat Gestures

First, we found that beat gestures (both when spontaneous or in a gesture-encouraged condition) were associated with significantly lower originality scores in individuals with high mental imagery skills. One possible explanation is that, although beat gestures are thought to facilitate verbal access and speech fluency, their production in individuals with high mental imagery may indicate an attentional narrowing or cognitive fixation during creative idea generation. This is in line with previous studies showing that beat gestures boost attention (e.g., Dimitrova et al., 2016; Wang & Chu, 2013). For example, Wang and Chu (2013) showed that beat gestures elicited smaller N400 components. In general, the N400 amplitude is smaller when it is easier to integrate the semantic meaning of a word into a context. The authors also attributed this finding to beat gestures’ role in attracting attention to focus words and emphasizing them. In the present study, beat gestures might have emphasized conventional uses, thus not allowing mind-wandering that could have led to more unusual solutions. A similar finding, but for iconic gestures and creative flexibility, was reported by Hyusein and Göksun (2024), who found that an increased use of spontaneous iconic gestures was associated with reduced creative flexibility. The authors interpreted this as a potential fixation effect of iconic gestures due to their imagistic representational nature.

Recent work on perception by Ongchoco and Scholl (2021) showed that regular auditory beats can scaffold attention. Scaffolded attention is a perceptual phenomenon where individuals experience “everyday hallucinations” of structured patterns, such as seeing lines or letters when looking at a regular grid, such as graph paper. The auditory beats studied by Ongchoco and Scholl (2021) were identical metronome clicks, such as click-click-click…, but our brains would automatically assign a structure to them, e.g., grouping them into units, such as ONE-two-three-ONE-two-three… Beat gestures, similarly, might act as scaffolds for visual attention and structure the cognitive and verbal rhythm during creative problem-solving. When mental imagery is high, it might be hard to disengage from the visual rhythm of the initial ideas and explore new ones.

Moreover, beat gestures might also indicate overconfidence. For instance, Ferrari and Hagoort (2024) found that beat gestures caused the illusion of high speaker confidence among observers. It might be the case that beat gestures also influence the speaker’s metacognition by signaling self-confidence. This is also in line with Yılmaz et al.’s (2025) findings that the use of non-representational gestures when speech is disfluent leads to higher task performance confidence. Similarly, in the current study, the use of beat gestures might have led the speaker to believe they were generating creative enough solutions, which might have discouraged them from seeking more novel alternatives. Alternatively, speakers might have used a lot of beat gestures to “mask” their lack of original ideas. For example, when there are speech disfluencies, speakers tend to use beat gestures to signal to the listener that they are having difficulty finding the right words (Kısa et al., 2024). In our case, it was finding the right ideas; but why did only people with high mental imagery do that? It could be because they had better metacognitive skills, as previous research has shown that training mental imagery could increase metacognitive skills (Rademaker & Pearson, 2012). High mental imagery individuals might simply be more aware of their performance and make an effort to signal such a difficulty to their interlocutor. Further research is needed to determine the exact role (e.g., self- or listener-oriented) of beat gestures in people with different levels of mental imagery.

Unlike Experiment 1, which demonstrated a significant interaction between beat gestures and mental imagery for originality in divergent thinking, none of the models in Experiment 2 yielded significant main or interaction effects for beat gestures or imagery skills during the AUT. A post hoc power analysis indicated that the current sample was slightly underpowered (power = 0.69), which may have limited our ability to detect subtle interaction effects. However, visual inspection of the interaction plots did not suggest even a trend toward an interaction, weakening the case for a false negative result. This points to the possibility that the original effect may not replicate in the absence of gesture encouragement or may be context-specific. For example, in Experiment 1, participants had one minute to solve the task, while in Experiment 2, they had three minutes. Studies have shown that as time passes by, the creative quality of ideas increases, and people are more likely to come up with more original ideas later on after the common ideas have been exhausted (George & Wiley, 2020). It might be the case that in the initial stages, when there are fewer original ideas, beat gestures have a greater impact on idea quality, but as time passes by, these effects wear off.

In contrast, for convergent thinking (RAT performance), we found a significant interaction between beat gesture frequency and vividness of mental imagery during the RAT. Beat gestures were associated with lower RAT scores at low and average levels of vividness but not at high levels. This replicates and extends findings from Hyusein and Göksun (2023), who reported that encouraged beat gestures hindered performance in individuals with low imagery ability. The current data suggest that this effect also occurs when gestures are produced spontaneously and is moderated specifically by the experienced vividness of mental imagery during the task, rather than general imagery ability. Interestingly, the direction of the interaction contrasted with earlier findings on representational gestures, which were shown to impair convergent thinking in high-imagery individuals (Hyusein & Göksun, 2025). That was, however, data for L1 Turkish and L2 English combined, and interaction plots showed that this effect was driven primarily by the L2 English condition, even though the interaction was not significant. It might be that there are different mechanisms occurring in L2, and the role of representational and non-representational might differ. The current findings reveal that mental imagery vividness and skills might be necessary for beat gestures to help, or at least to be positively associated with convergent thinking.

Beat gestures seem to benefit from mental imagery for convergent thinking, but high mental imagery in addition to gesture use seems to hurt or to have an inverse relationship with divergent thinking. This finding seems plausible given that convergent thinking requires more focused attention, i.e., finding the single correct solution, and divergent thinking benefits from leaky attention, i.e., exploring many alternative paths of solution. However, Zabelina et al. (2016) found that it was real-world creative achievement that was associated with leaky attention, while laboratory tests of divergent thinking, as is the case for the AUT, required more selective yet flexible attention. In our study, increased mental imagery and beat gesture use might have hindered attentional flexibility.

Finally, we detected a negative association between spontaneous beat gestures and vividness of mental imagery during divergent thinking. This negative correlation between beats and imagery vividness also supports the literature claiming that beat gestures are less imagistic (e.g., Hostetter & Alibali, 2007, 2019; Kita et al., 2017; McNeill, 1992)—in the presence of vivid mental imagery, their use is reduced.

5.2. Palm-Revealing Gestures

Regarding palm-revealing gestures, the between-group comparison in Experiment 1 revealed a significant negative association between palm-revealing gesture frequency and the flexibility dimension of divergent thinking in Group 2. There was also a significant main effect of palm-revealing gestures on elaboration, showing a negative association across both groups’ gesture-encouraged conditions. These findings suggest that when individuals are prompted to gesture while trying to solve a problem, palm-revealing gestures may act as bottom-up metacognitive cues that signal uncertainty or difficulty with the task (see Koriat et al., 2006; Yılmaz et al., 2025), which results in further uncertainty and a feeling-of-not-knowing. Even though encouraging gesture use introduces top-down (strategic) effort, palm-revealing gestures might still be interpreted or function as bottom-up indicators of struggle (because palm-revealing gestures are not explicitly and specifically encouraged); thus, they may interfere with the creative flow. In other words, palm-revealing gestures may disrupt creative flexibility by reinforcing a feeling of being cognitively stuck or disengaged.

Moreover, palm-revealing gestures during convergent thinking significantly interacted with mental imagery skills—they were negatively associated with RAT performance for low imagers. However, we did not observe a similar interaction when vividness was used as the moderator in this model. This dissociation may reflect subtle distinctions between stable mental imagery ability and task-specific imagery vividness as previously suggested (e.g., Hyusein & Göksun, 2025; McAvinue & Robertson, 2007).

We also found that responses on the AUT were more elaborate in the gesture-encouraged condition compared to the gesture-spontaneous one. This pattern occurred independently of gesture frequency, suggesting a practice or strategy adaptation effect (e.g., adding details to responses to fill in the time) rather than a gesture-specific benefit. However, findings in the same sample (Hyusein & Göksun, 2024) showed that iconic gestures were the only predictors of elaboration, not the gesture condition. Therefore, it seems to be an effect of iconic gestures. The more iconic gestures people produced, the more elaborate their answers were. Thus, it is representational gestures that signal elaboration, and non-representational gestures do not play a role in this aspect of creativity.

Together, these findings highlight that non-representational gestures, unlike representational ones, may not inherently facilitate divergent thinking. Instead, using those gestures, particularly in high-demand or effortful contexts, may serve as indicators of cognitive difficulty rather than creative enhancement. Furthermore, these results underscore the importance of considering individual cognitive differences, such as mental imagery skills, when interpreting gesture-related effects on creativity.

5.3. Theoretical Implications and Future Directions

Theoretically, these results support the view that representational gestures, probably due to their imagistic properties and semantic integration with the speech content, are more helpful for the speaker’s cognition compared to non-representational gestures. Non-representational gestures seem to influence speakers’ thinking in a more detrimental manner, either by implying overconfidence (e.g., see Yılmaz et al., 2025) or increasing cognitive rigidity (e.g., Dimitrova et al., 2016; Ongchoco & Scholl, 2021; Wang & Chu, 2013), as beat gestures did during divergent thinking, or causing a perception of not-knowing, e.g., palm-revealing gestures (Koriat et al., 2006). As we did not directly manipulate gestures (e.g., by instructing participants to use beat or palm-revealing gestures specifically) but encouraged overall use of the hands when speaking and thinking, it is difficult to infer a cause-and-effect relationship. Future research could study the temporal alignment of speech and gestures. For example, if creative thinking started deteriorating (for instance, ideas became less original or the RAT solution rates dropped) after participants started using non-representational gestures, then we might be able to attribute diminished creativity to the use of non-representational gestures. Unlike representational gestures, non-representational gestures might act as metacognitive cues. Recently, a framework of creative metacognition (CMC) has been proposed by Lebuda and Benedek (2023) and empirically tested (Lebuda & Benedek, 2024). This framework suggests that metacognitive knowledge, monitoring, and control in creative performance could enhance creative cognition. On the other hand, having the metacognitive knowledge of not-knowing, expressed by palm-revealing gestures, might diminish the speaker’s confidence and have a detrimental effect on idea generation. Future research could specifically manipulate non-representational gesture use to test if they might be related to creative metacognition and whether they could facilitate or impair it. Spontaneous gestures and encouraging gestures without explicitly specifying the type of gestures to be used might be insufficient to infer a causal effect of gestures on creativity.

6. Conclusions

In conclusion, the present findings suggest that non-representational gestures, such as beat and palm-revealing gestures, do not consistently support creative thinking and may, under certain conditions, hinder it. Rather than facilitating idea generation or insight, these gestures appear to reflect underlying cognitive states—such as fixation, overconfidence, or uncertainty—that can interfere with performance, particularly when mental imagery is low or when tasks are demanding. Beat gestures, for instance, may narrow attentional focus or elicit overconfidence, leading to less original idea generation or impaired insight. Palm-revealing gestures, on the other hand, appear to function more as bottom-up metacognitive signals of uncertainty, particularly in cognitively demanding or unfamiliar contexts. Critically, the effects of these gestures were not uniform but varied by gesture type, task context (divergent vs. convergent thinking), and individual differences in mental imagery.

These findings underscore the need to differentiate between gesture types in models of embodied cognition and to consider gestures not only as facilitators of thought but also as reflective indicators of internal cognitive states. Future work should examine the causal role of gesture timing and function, especially through experimental manipulations of gesture type, and should explore how non-representational gestures might intersect with emerging models of creative metacognition.