Actions chains and intention understanding in 3- to 6-year-old children

Significance

We examined here whether preschoolers (children aged 3 to 6 y) are able to organize intentionally their motor acts and have the capacity to understand others’ intentions by observation. We show that preschoolers, in spite of their young age, organize their motor acts intentionally and understand, although with a delayed ability with respect to older children, the intentions of others by action observation. Our study also highlights nuances related to the development of these skills that continue refining with age.

Abstract

In intentional behavior, the final goal of an action is crucial in determining the entire sequence of motor acts. Neurons have been described in the inferior parietal lobule of monkeys, which besides encoding a specific motor act (e.g., grasping), have their discharge modulated by the final goal of the intended action (e.g., grasping-to-eat). Many of these “action-constrained” neurons have mirror properties responding to the observation of the motor act they encode, provided that this is embedded in a specific action. Thanks to this mechanism, the observers have an internal copy of the whole action before its execution and may, in this way, understand the agent’s intention. The chained organization of motor acts has been demonstrated in schoolchildren. Here, we examined whether this organization is already present in very young children. To this purpose, we recorded EMG from the mylohyoid (MH) muscle in the children aged 3 to 6 y. The results showed that preschoolers, like older children, possess the chained organization of motor acts in execution. Interestingly, in comparison to older children, they have a delayed ability to use this mechanism to infer others’ intentions by observation. Finally, we found a significant negative association between the children’s age and the activation of the MH muscle during the grasp-to-eat phase in the observation condition. We, tentatively, interpreted it as a sign of an immature control of motor acts.

The “chained” organization of motor acts is a fundamental aspect of action organization of primates. This type of organization was first described in the inferior parietal lobule (IPL) of the macaque monkey by Fogassi et al. (1). The authors found that in IPL, there is a set of neurons that they named action-constrained neurons, which encode a specific motor act (e.g., grasping). Their defining feature is that their discharge is modulated by the action in which the motor act is included (e.g., grasping-to-eat, grasping-to-place, etc.) (1–3). It has been proposed that this “chained” organization of motor acts underlies a fundamental aspect of action execution: fluidity. Indeed, fluidity requires a close link between the motor acts forming the whole action so that its execution can occur without any interruption.

A large number of IPL-grasping neurons show mirror properties (see ref. 4), firing both during the execution of a motor act as well as during the observation of the same motor act. Most interestingly, in many action-constrained neurons with mirror property, the discharge intensity during action observation was modulated by the action in which the motor act was embedded. In this way, the activation of IPL action-constrained mirror neurons provides information not only on the observed motor act (e.g., grasping) but also predicts the final goal of the observed action (i.e., grasping-to-eat or grasping-to-place) allowing one to infer the agent’s intention (1, 5).

There is strong, albeit indirect, evidence, that a similar motor-act chained organization is also present in humans. Cattaneo et al. (5) recorded the electromyography (EMG) from the mylohyoid (MH) muscle, a muscle involved in mouth opening, in children aged 6 to 9 y. Children were instructed to grasp a piece of food or a piece of paper and to bring it to the mouth (grasping-to-eat) or to place it into a container (grasping-to-place), respectively. The results showed consistently greater activation of the muscles involved in mouth opening during the grasping-to-eat action, relative to the grasping-to-place one. Furthermore, the MH muscle activation started very early—already during the reaching phase—in the grasping-to-eat action relative to the grasping-to-place action. A clear difference in the MH activation between grasping-to-eat and grasping-to-place actions was also found when children observed the experimenter performing those same actions. As in action execution, also during action observation, the activation of the MH was greater and started earlier in the grasping-to-eat than in grasping-to-place action. Thus, the action-chain mechanism appears to be involved in the selection of motor acts for both action execution and understanding the intentions of others.

Based on these findings, in the present study, we examined whether preschoolers aged 3 to 6 y are able to organize intentionally their actions and to understand the agent’s motor intentions behind the observed actions. To this aim, the activation of MH muscle was recorded while children were either executing or observing two actions: 1) reach out to pick up a piece of food with the aim of eating it and 2) reach out to pick a piece of paper with the aim of placing it into a container. Note that both eating and placing actions develop quite early in infancy. By the time children are 1 to 2 y old, they usually have developed the ability to pick up food with their fingers, use a spoon or fork, and start feeding themselves. Infants begin to explore their environment and manipulate objects in the first year of life and may engage in simple actions like placing objects in their mouths. These actions are typically exploratory rather than purposeful. Toddlers (1 to 3 y) start to develop more intentional actions related to placing objects. They may begin to stack blocks or place toys in containers. This stage is marked by trial and error as they learn about spatial relationships and cause-effect. During preschool years (3 to 6 y), the development of children’s fine motor skills enables them to become good at placing objects precisely (6–9).

The results of the present study showed that preschoolers were able to modulate their intentional motor chain during action execution. However, unlike older children (5), they showed a reduced ability to use this mechanism to infer others’ intentions. Finally, we found a significant negative association between the children’s age and the activation of the MH muscle during the grasp-to-eat phase in the observation condition. We interpreted the stronger MH activation in younger children, relative to the older ones, as a sign of weak, immature control motor acts.

Results

Main Analysis.

The activity of the MH muscle was recorded in 18 (n = 18) neurotypical preschool children during the execution and the observation of grasping-to-eat and grasping-to-place actions. In the execution condition, the children actively performed two actions. Namely, they grasped the food and brought it to their mouth to eat (grasping-to-eat execution condition) or placed the piece of paper into the container positioned on their shoulder (grasping-to-place, execution condition). In the observation condition, children were instructed to observe carefully the experimenter while the experimenter was grasping a piece of food located on a touch-sensitive plate, bringing it to the mouth, and eating it (grasping-to-eat observation condition), or while the same experimenter was grasping a piece of paper from the same touch-sensitive plate and placing it into a container positioned on the experimenter’s right shoulder (grasping-to-place observation condition).

In both execution and observation conditions, the children’s MH activity was recorded using surface electrodes. Two devices, a button and a touch-sensitive plate, were used to signal the release of the hand from the table and the contact of the hand with the stimuli, respectively. These signals enable the subdivision of the observation and execution tasks into three epochs (reaching, grasping, and bringing, see Materials and Methods).

For each child, the EMG of MH muscle was recorded, rectified, and averaged separately in the three epochs (see above) for the two actions (grasping-to-eat and grasping-to-place) and the two conditions (observation and execution) and used as the dependent variable in the analyses. Two General Linear Model (GLM) analyses were then carried out to assess an anticipation effect of object grasping independently for each condition (execution, observation), each one with two levels of action type (eating and placing) and three levels of epoch (reaching, grasping, and bringing) as within-subject factors. Post hoc comparisons were Bonferroni corrected and the Greenhouse–Geisser (G-G) correction was used when the sphericity assumption was violated.

MH Muscle Activation during Action Execution.

As shown in Fig. 1, Right, the results revealed a main effect of action type, F(1, 10) = 18.16, P < 0.001, partial-η2 = 0.65, δ = 0.97, showing that the increase in MH activation was greater during eating action compared to placing action. In particular, there was a significant difference in MH muscle activation between the execution of eating and the execution of placing already during the reaching epoch, Mdiff = 7.01, SE = 2.76, P < 0.05. This increase became more pronounced during the grasping epoch, Mdiff = 7.23, SE = 2.36, P < 0.01, and reached its peak during the bringing epoch, Mdiff = 11.67, SE = 2.40, P < 0.001. These findings are further supported by the main effect of epoch, F(2, 20) = 9.85, P < 0.001, partial-η2 = 0.50, δ = 0.97, indicating that independent of action type, MH activation was significantly higher in the grasping epoch compared to the reaching epoch, Mdiff = 2.81, SE = 0.77, P < 0.01, as well as in the bringing epoch compared to the reaching one, Mdiff = 4.63, SE = 1.06, P < 0.01.

Fig. 1.

Left: Time course of the rectified EMG activity of the MH muscle during the execution of the grasp-to-eat action (blue) and the grasp-to-place actions (orange). The curves are aligned to the moment of stimulus lifting from the touch-sensitive plate (T0). The pale green box outlines the phase of the reaching movement; the dashed vertical line indicates the grasping phase; the yellow box outlines the phase of the bringing movement. Right: mean EMG signal amplitude of the MH muscle during the eating and placing actions in the execution condition for the three phases (reach, grasp, and bring). The bars represent the SEM. * indicates significant differences (P < 0.05).

In Fig. 1, Left, shows the time course of the median EMG signal of the MH muscle in the two conditions (grasping-to-eat and grasping-to-place). In the eating condition, the activation of the children’s MH muscle started to increase several milliseconds before the hand grasped the food (T0), continued to rise during the grasping phase, and as expected, reached its peak toward the end of the action.

MH Muscle Activation during Action Observation.

As shown in Fig. 2, Right, the results revealed a main effect of action type, F(1, 13) = 19.68, P < 0.001, partial-η2 = 0.60, δ = 0.98, showing that MH activation was greater during the eating action compared to the placing action. Additionally, a main effect of epoch, F(2, 26) = 34.04, P < 0.001, partial-η2 = 0.72, δ = 1, indicated a greater activation during grasping compared to both reaching, Mdiff = 3.33, SE = 0.45, P < 0.001, and bringing epochs, Mdiff = 3.68, SE = 0.62, P < 0.001. Finally, the results showed a significant interaction between action-type and epoch, F(2, 26) = 4.82, P < 0.05, partial-η2 = 0.27, δ = 0.55. Post hoc comparisons showed that the interaction effect stems from a significant difference in MH activation between eating and placing actions both during the grasping epoch, Mdiff = 3.26, SE = 1.07, P < 0.01, and the bringing epoch, Mdiff = 1.17, SE = 0.29, P < 0.01. No significant difference was found between eating and placing actions during the reaching epoch, Mdiff = 0.45, SE = 0.25, P = 0.09. To further examine the null effect, an equivalence test [Jamovi, v.2.3.21.0; TOSTER 0.4.0 (10)] revealed no significant difference between the two experimental conditions (eating vs. placing) for the variable “reach” (t = −2.01, df = 14, P = 0.07). The equivalence intervals were [0.15, −4.16] (SI Appendix, Fig. S1 and Equivalence Test). A sensitivity analysis further supported the robustness of the effects given our sample size (SI Appendix, Sensitivity Analysis).

Fig. 2.

Left: Time course of the rectified EMG activity of the MH muscle during the observation of the grasp-to-eat action (blue) and the grasp-to-place (orange) actions. The curves are aligned to the moment of stimulus lifting from the touch-sensitive plate (T0). The pale green box outlines the phase of the reaching movement; the dashed vertical line indicates the grasping phase; the yellow box outlines the phase of the bringing movement. Right: mean EMG signal amplitude of the MH muscle during the eating and placing actions in the observation condition for the three phases (reach, grasp, and bring). The bars represent the SEM. * indicates significant differences (P < 0.05).

The time course of the activations in the two conditions is shown in Fig. 2, Left. In both grasping-to-eat and grasping-to-place actions, the activation of the MH muscle slightly increased during the reaching epoch, peaked when the child observed the experimenter grasping the food (T0), and subsequently decreased.

Age Correlations: Exploratory Analysis.

To investigate whether MH muscle activation changed as a function of the children’s age, Spearman correlations (one-tail) were examined between children’s age (months) and the activation of EMG of the MH muscle in conditions (observation, execution), action types (eating, placing), and at the three epochs (reaching, grasping, bringing). The results showed a negative correlation between the children’s age and the activation of the MH muscle during the grasp-to-eat phase in the observation condition, rho(16) = −0.56, P < 0.05 (Fig. 3). Bootstrap analysis, based on 1,000 resamples, supported the stability of the correlation coefficient, with a bias of 0.02 and a SE of 0.19. The 95% CI ranged from −0.824 to −0.128. These findings suggest a consistent negative association between participant age and MH activation, with greater activation found in younger children.

Fig. 3.

Spearman correlations between children’s age and rectified EMG activation of MH muscle during the observation of the grasp-to-eat action (T0).

Discussion

The ability to organize sequentially motor acts according to the final goal of an action is a key step in achieving the fluidity of actions typical of adult motor behavior. Similarly, the ability to understand the final goal of others’ actions is vital for the development of social competencies. As far as these abilities are concerned, there is ample evidence that both are mediated by the mirror mechanism (1, 11–14). This mechanism transforms, in the observer’s brain, sensory representations of others' behavior into motor representations of the same behavior, thus allowing the observer to understand the goal of the motor acts as well as, through the “action constrained” mirror neurons (1, 2), the agent’s intention.

In this study, we examined whether the ability to organize actions intentionally and understand others’ action intention, i.e., the final goal behind the observed actions, is already present at preschool age. To this purpose, the activation of MH muscle was recorded in 3-to-6-y-old children, while they were executing or observing two actions: 1) reaching to grasp a piece of food with the aim of eating it and 2) reaching to grasp a piece of paper with the aim of placing it into a container. The results showed that preschoolers are able to select the appropriate motor chain during action execution. Specifically, during the execution of the grasping-to-eat action, a significant increase in the activation of the MH muscle was found already during the reaching epoch, several milliseconds before the child actually grasped the food. The activation of the muscle persisted during the grasping epoch and peaked when the children opened their mouth. No similar activations were found during grasping-to-place action. These results appear to reflect a physiological mechanism whereby, during the preparation of a goal-directed action, there is activation of the corresponding motor chain before the start of the actual movement, being the selection of the “chain” driven by the intention to perform a given action and to reach a specific goal. These findings are in line with previous results reported in older children (6 to 9 y old) (5), and, most importantly, demonstrate that the ability to select intentionally sequential motor acts is already present in preschool children.

Unlike in action execution, during the observation of grasping-to-eat, the chain mechanism was activated slightly later, relative to older children (5), and, specifically, only when the experimenter initiated the actual grasping movement. Indeed, no substantial difference was found between the eating and placing actions in the reaching epoch.

In this respect, the activation of the MH muscle in children aged 6 to 9 y in Cattaneo et al. (5) was already present 100 ms before object grasping, and the activity in the reaching epoch was significantly higher in the grasping-to-eat than in grasping-to-place conditions (5). These data indicate that while older children are able to both organize their actions intentionally (execution) and to understand the intentions of others (observation), preschoolers show a quite delayed activation during observation, most likely because their intention coding mirror mechanism is not sufficiently developed for a fast discrimination between others’ intentions.

As for this finding, it is worth comparing the behavior of typically developing (TD) preschoolers in the present study with that of children with autistic spectrum disorder (ASD) tested in Cattaneo et al. (5), who used the same paradigm as the present study. The results showed a complete lack of EMG activation in the MH muscle during the reaching epoch of grasping-to-eat action execution. Likewise, the children with ASD did not exhibit any activation of the MH muscle while observing the experimenter grasp the food to eat and, as a result, were unable to understand the intentions of others. Thus, our study found that preschoolers have a functioning anticipatory motor mechanism that underlies the sequential organization of actions. However, they exhibit a delayed ability to use their intentional mechanism to understand the intentions of others. In the present article, the expression “understanding actions of others” is used several times. But what does it mean exactly? The first answer that might come to mind derives from philosophical studies. It implies that the person observing the action of another person has some knowledge of the beliefs, desires, and intentions of the person executing that action. This knowledge allows the observer to understand the reasons underpinning the observed behavior. There are studies showing that the reasons underlying a given behavior are encoded in higher-order cognitive areas (see refs. 15 and 16). However, if we take a closer look at the commonly accepted meaning of “to understand,” we see that it does not exclude the possibility of understanding an observed action without necessarily having a knowledge of the mental states that have motivated it. Imagine, for example, a person reaching for a glass of beer and someone asking you if you have understood what that person is doing. Your answer would most likely be “This person is picking up that glass.” You might even say, “This person is going to pick up the glass to drink.” Your answer is most likely correct, even if you are not aware of the reasons that motivated the person to reach for the glass. In this case, understanding the action means to identify the goal of the motor act (pick up the glass) and to infer the possible motor intention underlying it (to drink) but not the reasons underlying this behavior (thirsty, drunkard, etc.). In order to clarify the meaning of “understanding,” Rizzolatti and Sinigaglia (13) proposed to call full understanding of the action if the understanding includes the reasons underlying the action, and the basic understanding of the action if the observer identifies the goal of the motor act and infers the intention behind the whole action without necessarily knowing the reasons that led to its execution (17). The action understanding described in Cattaneo et al. (5) is a basic understanding likely mediated by “action-constrained” mirror neurons (1, 5). Consistently, in a study by Boria et al. (18), ASD and TD children were presented with hand actions on a computer screen and asked to determine both the “what” and the “why” of the actions. For instance, the children were shown a pair of scissors being grasped by a hand. The grasping hand could be positioned on the scissors handle, as one does for using the scissors, or on the center of the scissors, as one does for moving them. Both groups of children responded correctly to the “what” question, stating that the hand was grasping an object; however, the ASD children failed to indicate the “why” of the observed gesture, most of the errors stemming from the “move” condition, i.e., the one showing an unusual action associated with the use of scissors. In a further experiment, the same two groups of children saw pictures showing a hand grasping an object as in the previous experiment but, this time, within a context suggesting either the typical use of the object or its explicit placement into a container. Here, children with ASD performed like TD children correctly indicating the agent’s intention. Therefore, based on the studies described above and our findings, it can be concluded that understanding the “what” of a motor act is a very early and robust acquisition. However, understanding the intention of an action, even a basic one, is a later acquisition that depends on the context in which the motor act is performed and, later, on the mirror intentional chains.

Another interesting preliminary finding of the present study—that would need further support given the limited sample size for the correlational analysis—is that the activation of the MH muscle at the actual grasping time of the stimulus during the observation of the experimenter’s grasping-to-eat action was greater in the youngest children. Beyond understanding actions, mirror neurons facilitate imitation (19), a well-known phenomenon in developmental psychology that allows children to learn from and interact with the world around them (20, 21). The increase of the MH muscle found in our sample during the observation of the grasping action in the youngest children is likely due to an immature frontal inhibitory control, which develops with age until early adulthood (22–24). This is consistent with our findings on the grasp-to-place conditions. The observed, albeit low, activation of the MH muscle during both the execution and observation of the placing action would reinforce the idea that preschoolers have not yet developed an efficient frontal inhibitory control by the prefrontal lobe on the mirror neurons coding motor acts.

In conclusion, taken together, the current data bring further support to the different stages of child development, in continuation of infant research showing an early attunement to goal-directed motor acts. This would enable very young children to anticipate the “what” of a motor act, such as the goal of “picking up the glass,” as referenced in the above example and supported by a robust body of evidence (25–29). Conversely, children’s understanding of others' motor intentions, the “why,” i.e., “to drink,” would develop more gradually, in line with the development of more sophisticated cognitive processes (30).

Materials and Methods

Participants.

The sample size was estimated using G-power with the following parameters for within-factors repeated measures: effect estimate = 0.3; probability error = 0.05; power estimate = 0.9; number of groups/conditions = 6; number of measurements = 20. The estimated sample size was 12 participants. In order not to exclude any children who, with parental consent, wished to participate in the study, all volunteers were examined. This also made it possible to take into account a conceivable percentage of dropouts or extreme cases. Eighteen (18) preschool neuro-TD children were then recruited from two kindergartens in Milan and Trecate, Italy (40% female; mean age = 4.31 y; SD = 0.76; age range = 3.08 to 6.01 y). Informed consent was obtained from all children’s legal guardian(s) in line with the Declaration of Helsinki and its revisions, as well as in accordance with the requirements of the ethics committee, Committee of the Department of Psychology (CERPS), Università Cattolica del Sacro Cuore, Milan, which approved the study.

Procedure.

We investigated the motor activation of the mouth-opening MH muscle during two different actions, i.e., grasping to eat and grasping to place, in two experimental conditions: observation and execution. Children were assessed in both experimental conditions. In the eating-observation condition, children were instructed to carefully observe the researcher using her right hand to grasp a piece of food from a touch-sensitive plate, bringing it to her mouth, and eating it. In the placing-observation condition, children observed the researcher using her right hand to grasp a piece of paper from the touch-sensitive plate and place it into a container situated on the experimenter’s right shoulder. Both actions were repeated 30 times. In the execution condition, children carried out the actions of grasping a piece of food to eat and a piece of paper to place into the container located on the child’s shoulder. Children repeated the actions 30 times.

Children were seated at a table in front of the researcher and, according to experimental conditions, all trials started with the experimenter or participant’s hand resting on the start button, and throughout the duration of the experiment, the activation of the participant’s MH muscle was recorded using surface electrodes. Prior to the experiments, children underwent a training session where they performed grasping and eating the food or placing the piece of paper into the container, depending on the presented stimulus. A brief training session was also conducted for the observation condition. The conditions were presented in a random order, blocked by trial type, namely, “observe placing action,” “observe eating action,” “execute placing action,” and “execute eating action.” This design allowed for a balanced and randomized distribution of trials across participants. No verbal instructions were provided during the trials, except for occasional prompts such as “Please pay attention” or “Look at the experimenter.”

To mark on the EMG recordings, the beginning and the end of each action and the contact with the stimuli, a button and a touch-sensitive plate were placed on the table and connected to the EMG apparatus. The release of the hand from the button signaled the beginning of the action, and pushing the button again indicated the end of the action. The grasping of the piece of paper or the food was detected through the touch-sensitive plate and marked as the starting moment (T0).

Materials.

For the recordings of the mouth-opening MH muscle, we used a wireless electromyograph (EMG; DueLite, OT Bioelettronica Srl, Turin, Italy). The MH muscle was recorded using two surface bipolar electrodes (CDE—Bipolar electrodes diameter 15 mm with concentric connector). The two electrodes were placed under the child’s chin for MH muscle recording. EMG was recorded continuously throughout the experiment. The interval between trials was according to the compliance of children. The signal was amplified 1,000×, sample frequency of 2,048 Hz, controlled by the EMG signal software, and stored for offline filtering (bandpass: 10 to 500 Hz) and further analysis.

Signal Processing and Definition of Epochs.

Trials were immediately discarded if the participant coughed, spoke, laughed, etc. Additionally, trials were discarded if MH values deviated more than two SD from the mean of the signal median recorded in each trial (SI Appendix, Table S1).

In all conditions, the start button and the touch-sensitive plate on which the food/paper was placed, were connected to the EMG apparatus. The actions were thus divided into three epochs: reaching the stimulus, grasping it, and bringing it to the mouth or into the container. The release of the start button signaled the beginning of the reaching epoch that ended just before touching the touch-sensitive plate. The contact with the plate on which the stimulus was put signaled the grasping epoch (T0). The release from the plate signaled the end of the grasping and the start of the bringing epoch. Video recordings were used to assess the time at which the hand reached the final target of the action (mouth or container, depending on action type). The average duration of the epoch for the children (execution) and the experimenter (observation) is shown in SI Appendix, Table S2. All recordings were aligned on the phase of paper or food lifting from the touch-sensitive plate (T0).

All recordings were filtered (30 to 400 Hz). The instants of signal activation were identified in each device’s recordings. The data points of the averaged recordings were then divided into bins of 200 ms.

Data Analysis.

Two GLM analyses were carried out using the EMG activation in the three epochs of action as a dependent variable. First, for each child in each condition, the median of each trial was calculated, and the SD (−2, +2) was used to normalize the data. As a result of this procedure, some children were automatically excluded by the system from some conditions. Consequently, the data analyses (GLMs—see below) considered only those cases that were retained in all conditions. In both conditions (execution, observation), within-subjects ANOVAs for repeated measures were carried out with two levels of action type (eating, placing) and three levels of epochs (reaching, grasping, and bringing).

Post hoc analyses were Bonferroni corrected. Significance levels were set at P < 0.05. A G-G correction was applied whenever the sphericity assumption was violated. Additionally, correlation analyses (Spearman rho) were carried out to explore the relationship between the MH muscle activation and children’s age.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Anonymized EMG data have been deposited in Zenodo Archive (https://zenodo.org/records/11918084) (31).