Social interactions are one of the core components of our daily lives. Neurotypical infants make attempts to engage in social interaction from as early as 6–8 weeks of age, responding to their caregivers by smiling or cooing [1]. Children spend longer looking at faces than at comparable non-social stimuli with matched spatial frequency [2], they prefer biological to non-biological motion [3], and smile more in response to social than non-social stimuli [4]. These studies indicate high social reward responsivity in early years of life. This bias for social stimuli is also shaped by contextual parameters such as familiarity[5].

While there are several paradigms to evaluate social reward responsivity (47), we focus on two most commonly used classes of paradigms used to study this construct. One class consists of passive paradigms that present the social stimulus (e.g. a smiling face) and measure the pleasure/consummatory aspect of the participant’s response. These measures can be observational (e.g. gaze duration in an eye-tracking measure, or reward-related neural response in a neuroimaging measure), or self-reported (e.g. ‘rate how much you like this’). In contrast, the second class of paradigms are instrumental tasks that present conventionally rewarding social stimuli contingent on the performance of a particular behaviour (typically a motor act such as pressing buttons). In these tasks, the choice behaviour as indexed through the number or speed of the responses is taken to provide a measure of social reward responsivity. While the former class of paradigms is related to how much children want and enjoy engaging in a social interaction, the latter is more related to whether social interaction can reinforce other behaviours. Both of these components are vital for social functioning.

While self-report studies arguably provide a subjective account of how rewarding a stimulus is, ratings and language-based tools are easily influenced by experimenter or subjective biases and are not suitable for use in populations with low verbal abilities, such as very young children [6–8]. To minimise such biases, preferential looking paradigms using camera-based or continuous eye-tracking technology have been used to index relative reward value [9, 10]. These studies are based on the assumption that longer gaze duration toward a specific set of stimuli indicates greater interest and greater positive value for those stimuli [10–12]. In the context of this study, we only discuss preferential looking paradigms as used to assess relative preference between different rewarding stimuli. In a typical preferential looking paradigm for social versus non-social rewards, two competing stimuli- one of each type- are simultaneously presented while participants’ gaze is tracked [13, 14]. These studies have demonstrated that neurotypical children and adults gaze longer at social than non-social stimuli [14–17]. Children as well as adults thus demonstrate a robust preference for social rewards, when measured using visual preference paradigms.

Comparatively fewer studies have used an instrumental task to test preference for social stimuli. One of the early attempts at addressing this gap was the Social Incentive Delay task (SID), in which participants are aware of the reward (social/non-social) they would receive at the end of each trial on a reaction time game [18]. Here, reaction time in anticipation of social reward is used as a measure of social reward responsivity. Neurotypical adults react faster to monetary rewards than social (smiling faces) rewards [18]. Another study using a similar paradigm compared two age groups of younger adults (20–28 years) and older adults (60–78 years), both of whose hit rates (accuracies), were significantly higher for monetary than social (smiling faces) rewards [18, 19]. The same study found that the reaction time was influenced by the reward magnitude and not the type (social vs. monetary), thus providing no evidence of preference for social stimuli. The second set of studies examining social reward responsivity in instrumental paradigms used effort as an index of social reward responsivity. In tasks of this type, participants exert different amounts of effort (made through button presses) to obtain a social or a non-social reward [20, 21]. These tasks demonstrate a preference for social stimuli, in as young children and neurotypical adults invest more effort to get social reward than non-social rewards. However, no such preference is seen in neurotypical adolescents [21–23]. Ruta and colleagues used a similar but simpler paradigm targeted for younger children (14–68 months), where children choose one of two buttons to look at social or non-social images [24]. They found that while neurotypical children do not show any preference for the ‘social’ stimulus, autistic children demonstrated a lower preference for social stimuli in the choice task. The authors interpreted the lack of a preference for social stimuli in neurotypical children to the nature of nonsocial stimuli used (trains) in this study. It was hypothesised that choosing a less salient nonsocial stimulus (e.g. a geometric pattern) might reveal a preference for social stimuli in neurotypical children.

In summary, studies evaluating social reward responsivity in the neurotypical population have reported mixed findings ranging from high preference for social stimuli to no preference. In paradigms where the stimuli are only passively processed, measuring consummatory behavior, participants typically show a higher preference for social stimuli. On the other hand, paradigms that involve an instrumental component do not always show this ‘social advantage’. The scope of inference has been limited by the absence of any study using both of these paradigms in the same set of individuals. The current study addresses this gap using both a preferential looking task and an instrumental choice task in young children administered on a mobile platform (a tablet PC). In line with previous lab-based studies, we hypothesise that children will show a visual preference for social stimuli in the both tasks.

An outlier check was done for both sets of data using Cook’s distance taking cut off as 4/n. This identified seven potential outliers for the preferential looking task and nine potential outliers for the button task. Statistical analyses were run without these participants for both the tasks. We found that exclusion of any of these data points did not change the findings, so we report the results with the full sample including all the participants.

This study evaluated two putative task measures of social reward responsivity in typically developing children using a tablet PC. We found a preference for social over non-social stimuli in a preferential looking task, replicating results from similar studies using lab-based eye-trackers [15, 16]. This preference for social stimuli was not seen in the instrumental choice (button) task. Despite this absence of overall preference for social stimuli in the button task, a positive relationship was noted between children’s social skills and their preference to choose the button associated with the social stimulus, replicating previous results. Interestingly, these two measures of preference for social stimuli were uncorrelated in this dataset, pointing to the possibility that these tasks index distinct processes within the broader construct of social reward responsivity. We discuss these results below, highlighting the importance of task choice in measuring social reward responsivity and its potential impact.

Studies evaluating visual preference for social stimuli in neurotypical children suggest that they spend longer looking at social stimuli [16, 17, 26, 37, 38]. It is also suggested that the more complex social stimuli (depicting multiple individuals), when used in a preferential looking task, are likely to elicit greater preference for social stimuli [39–41]. In the present study, we report similar results showing preference for social stimuli in neurotypical children using a tablet-based preferential looking task. A significant gender by age interaction effect was noted, with younger females showing greater preferential attention to social over non-social stimuli. This result mirrors previous findings of higher preference for social stimuli in females at both behavioural and neural levels [42, 43]. Similar effects of age (but not gender) on social reward responsivity has also been reported in another study suggesting stronger preference for social stimuli in younger (4–8 year old) participants compared to older children (9–12 year) [23].

Unlike preferential looking measures, behavioural approach or instrumental choice tasks have been less consistent in indicating preference for social stimuli in neurotypical children [21–24]. In the present study, we replicated the findings by Ruta and colleagues [24] using an adapted version of the original task. Children showed no significant preference for social stimuli on the button task. This observation suggests that the results by Ruta and colleagues cannot be explained fully by the choice of trains as the nonsocial stimuli. Interestingly however, a reduced preference for social stimulus was associated with low scores on a measure of social skills (SAS). Responses to social rewards have been suggested to be linked to social skills [30, 31]. This result is also consistent with the earlier study which found a reduced preference for social rewards in autistic children, who typically score low in SAS [29].

Even though the button task is a simpler version of the instrumental tasks used previously, it still requires participants to remember the association between the cue (button) and reward (linked videos). Furthermore, this task’s requirement to make motor movements to seek the social reward imposes additional demands. Thus, performance in the button task is likely to tap into processes beyond those engaged solely in reward processing. Preferential looking tasks, on the other hand, impose few/no additional demands and hence might be more directly related to consummatory or hedonic preference for social stimuli. While both these tasks seem to evaluate the common concept of social reward responsivity, it is possible that they may index distinct processes. Importantly, both the preferential looking as well as the button task for measuring social reward responsivity were administered on an overlapping sample of children, making it possible to test their inter-relationship. Interestingly, we did not find evidence for a significant relationship between the preference for social stimuli as quantified by these two tasks. While any interpretation of the lack of a relationship should be done with caution, this result supports the possibility that these two tasks tap into distinct aspects of social reward processing. Similar dissociations in different aspects of reward processing–such as wanting and liking, have been reported for other stimuli such as food or drugs, and these have been linked with atypical behaviour [44, 45]. Dissociations have also been reported between self-report and behavioural measures of social reward responsivity in clinical populations [46, 47]. Accounts of reduced social reward responsivity in clinical populations may therefore benefit from a more rigorous task-based assessment of putatively distinct components of reward processing in future studies.

Several limitations arise from the constraint of implementing these tasks in a new, mobile computing environment. First, the order of the two tasks was not counterbalanced: the preferential looking task was always run first, followed by the button task. However, it is unlikely that the fixed order of administering the two tasks would have influenced the results. Since the preferential looking task presented both social and nonsocial stimuli simultaneously, participants are not likely to experience selective fatigue with one of these two categories of stimuli. Accordingly, their behaviour on the button task is not likely to be biased toward social or nonsocial stimuli due to having completed the preferential looking task earlier. Second, the order of presentation of stimuli in the preferential looking task was non-random (social stimuli on the left in the first four trials and on the right in the next four trials). While this fixed order of counterbalanced stimuli should not influence the inferences drawn, we have fully randomised the order of presentation in a later study–and demonstrated similar results [35]. Third, in the instrumental task the association between button colour and stimulus type (social/non-social) was not counterbalanced across participants. Since the participants does not show any difference in the rates of pressing either button, it is unlikely to be a confound for the current study. However, we have added this feature in the later version of the task, to guard against confounds with baseline colour preferences [35]. Fourth, there was no memory check at the end of the button task to test if children remembered the association between the colour of the button and the type of stimulus (social/nonsocial). It is therefore possible that some participants may have forgotten the association, thus increasing noise in the data. Therefore, we recommend the use of a formal memory check at the end of similar tasks that rely on pairing between stimuli. Finally, while we administered both the preferential looking and the button task on all participants, technical errors resulted in a significant loss of data, resulting in >60% of the sample with data on both tasks. However, we find no reason to expect any of these limitations to change the interpretation. The sample of the children included in the study were recruited from an event held in the university for introducing scientific concepts to young children. It is possible that such events may be attended more by the children and families that are from higher educational backgrounds. It will be important to see if the findings can be replicated with a representative sample collected from the community.

The present study reports data from two tasks that evaluate two complementary aspects of social reward responsivity, in young children. The results highlight the importance of using a range of measures to tap these different components of social reward processing. These complementary measures of social reward responsivity can provide a theoretically informed approach to stratification within and across neurodevelopmental conditions.