The main finding of our study is the one related to the influence of emotional stimuli on temporal prediction ability. Accordingly, to our a priori hypothesis, our results show a specific decrease of RTs, in the single-interval task, when fearful stimuli were used, with respect to neutral stimuli. Differently, from our a priori hypothesis we also found a specific increase in RTs for joy stimuli in the random context (i.e., no-time condition) compared to neutral stimuli.
It’s important to note that the emotional valence of the stimuli used in the TP-E task was validated independently. Specifically, twenty-eight participants not included in the main study, evaluated the valence and arousal of the stimuli used. For valence, fear FACS were judged significantly unpleasant compared to neutral and joy ones. Furthermore, joy FACS displayed significantly higher valence ratings compared to fear and neutral stimuli.
The effect of emotion on temporal prediction ability
Predictive contexts: single-interval and rhythmic
Our finding concerning the motor response facilitation for fear stimuli, compared to neutral and joy, in the single interval context aligns with previous research suggesting that decreased RTs may reflect enhanced motor system readiness in the context of the processing of negative emotion29. As an example, we recently investigated the capacity to process emotional body postures in a three-choice categorization task29. Our findings showed significantly decreased RTs for pictures depicting fearful (i.e., negative) body postures when compared with happy (i.e., positive) or neutral postures. Furthermore, RTs to negative body postures were not significantly different from RT to negative scenes taken from the International Affective Picture System (IAPS), suggesting a general augmented processing speed and resource allocation for fearful stimuli belonging to different categories.
Negative emotional stimuli have been demonstrated to modify early components of event-related potentials already after 100 ms of stimulus initiation, indicating a quick allocation of attentional resources30,31. The “negativity bias” for negative emotional stimuli has been reported for brain regions involved in emotion processing (e.g., the amygdala, the orbitofrontal cortex, or the insula) but also for motor areas involved in motor representation and planning (e.g., premotor cortex, supplementary motor area, and parietal cortex), and the primary motor cortex32,33. Early activation of sensorimotor regions is assumed to be the neurophysiological correlate of the rapid motor response observed in response to negative emotional stimuli29,33. Notably, enhanced readiness of the motor system is supposed to represent just one of the corollaries of emotional adjustments in the processing of threatening situations, including physiological arousal and cognitive appraisal34,35. Moreover, the modulation of temporal perception is one of these emotional adjustments attributed to readiness for threatening situations in emotional contexts (for a review see4). Specifically, studies, using aversive or negative emotional stimuli (faces with an emotional expression of fear or anger), have pointed out that the exposure to these stimuli induces an overestimation of elapsed time in time judgment tasks5,36. This behavioral response has been explained considering the internal-clock model of interval timing37,38,39. This model claims the existence of different stages underlying the perception and the estimation of time, positing the existence of an internal “clock” in the brain, regulating the sense of time. Time representation originates in a pacemaker-accumulator system; the pacemaker emits pulses at a constant rate that are counted in the accumulator. The greater the number of accumulated pulses is, the longer the duration is judged to be4. In addition, the content of the accumulator is retained in working memory and compared with stored duration in long-term memory to build time judgment (memory and decisional stages)4,37,39,40. In the context of the relation between time and emotion, the idea is that the internal clock speeds up in emotional contexts representing threatening situations, thus causing more pulses to accumulate for the same physical unit of time4. When the clock runs faster and more pulses are accumulated, the duration is judged longer.
Here, even if we did not explicitly assess time judgment using ad hoc protocol, such as temporal discrimination task, the facilitation of RTs for fear in the single-interval context could be explained in the framework of the internal clock model. We can infer, from our result, that RT decrease in temporal prediction, in fear situations, could be linked to increased readiness that indeed occurs in threatening situations. Moreover, we can also hypothesize that modulation of temporal perception, in terms of an overestimation of elapsed time, may have facilitated a decrease of RT in a temporal prediction task. This latter hypothesis should be properly addressed in a study combining time judgment and temporal prediction. In favour of our hypothesis, the overestimation effect of anger and fear on the perception of time was supposed to result from an automatic process linked to dopamine activity which allows humans to anticipate an event by preparing them to act quickly4. The more rapidly time passes, the sooner humans are ready to act. This speculation is largely consistent with our findings of reduced RTs in temporal prediction when the fear stimuli were used.
It is noteworthy to underlie that RTs decrease with fear stimuli was observed only when temporal prediction was based on an interval representation (i.e., single-interval context) and not when the prediction was derived from a rhythmic stream. Hence, in rhythmic context, there was no significant difference between emotional (fear and joy) and non-emotional (neutral) stimuli, suggesting that when temporal regularity is strongly established the predictability of the task itself overrides any modulation by emotional valence or arousal. Indeed, this is the only temporal context where RTs for neutral stimuli were comparable to those for emotional stimuli.
One possible explanation could be related to the fact that single-interval-based predictions are strongly dependent on cognitive abilities, such as attention and memory41. Negative emotions have been demonstrated to influence to the greatest extent cognitive functions such as memory or attention with negative context benefitting from prioritized processes in memory, compared to neutral ones, richly encoded, and recollected42,43. Furthermore, studies, exploring the influence of emotion on memory for temporal information, have underlined that negative stimuli also enhance a greater episodic encoding and recall for the temporal information of those stimuli, compared to positive and neutral ones18,19. When confronted with angry or fearful faces, the acceleration in clock speed described above in temporal estimation is accompanied by increased attention to the duration dimension of those stimuli associated with the fear-provoking event (for a review see4). It could be inferred that our result about the improvement in predictive ability (i.e., decrease in RT) for fear in the single-interval context, compared to neutral and joy, could also reflect enhanced attention to the duration dimension of the stimuli and an improvement in mnemonic encoding and recall for temporal information for negative stimuli. It is important to emphasize that there was no statistically significant difference between joy and fear in the single-interval context, which aligns with previous studies suggesting that both positively and negatively valenced arousing stimuli typically lead to transient overestimations of temporal durations7. Notably, the magnitude of this effect is generally larger for negative stimuli44 .In this case (e.g., memory-based context), our finding supports the idea that fear holds a privileged position, compared to joy, in memory and attention, potentially enhancing temporal prediction abilities in this context.
The second possible explanation could rely on the different cortical-subcortical networks implicated in temporal prediction in the two different contexts (interval-based vs. rhythmic)9. A pivotal role of the cerebellum in interval-based predictions has also been shown by Breska and Ivry (2018)9 and supported by other works in patients with cerebellar degeneration45,46 and in healthy subjects using neuromodulation14. In the last years, the hypothesis of cerebellar involvement in emotional processes has been reinforced47. Using emotional pictures, researchers have demonstrated cerebellar activations following the presentation of aversive pictures48, with temporal hierarchical processing of arousal and valence. Specifically, in the cerebellum, arousal was firstly processed (occurring in both vermal (VI and VIIIa) and hemispheric (left Crus II) lobules) followed by valence and its interaction with arousal (occurring in left V and VI lobules and Crus I)49. Arousal, specifically, has been identified as the component underlying the increased readiness response to threatening situations, as reported in response to expressions or body postures expressing fear or anger. Thus, we can assume that a preferred fast activation of the cerebellum in orchestrating the response to emotional stimuli may have facilitated temporal prediction in single-based predictions, that is when the cerebellum is preferentially engaged. It is worth noting that these two explanations are not mutually exclusive, and both hypotheses may theoretically explain the selective decrease in RT in temporal prediction based on an interval representation (single interval) when fearful faces were utilized.
No-predictive context: random
An equally important result to consider is the specific increase in RTs observed for joy stimuli in the random context (i.e., no-time condition), compared to neutral.
Excluding the temporal properties of the context, since in the random context there was no temporal information to be implicitly acquired, a possible explanation could be related to the influence of emotional arousal on reaction times. Here, the preliminary evaluation of the stimuli used in the study showed that joy FACS were rated as significantly more arousing than both neutral and fear FACS. The relationship between arousal and motor performance has been described as a curvilinear relationship50,51. This model50 described that motor or cognitive performance improves with physiological or perceived arousal, reaching the optimal performance at an intermediate level of arousal. When arousal levels become excessively high, performance tends to decline. Likewise, we observed that RTs were significantly faster for neutral stimuli perceived with low to intermediate arousal levels compared to joy and fear. Noteworthy, in this study, we performed a subjective self-rating evaluation and to strengthen our hypothesis related to arousal influence on motor responses in no predictive context, it would be interesting, in future studies, to record physiological responses related to arousal (e.g., skin conductance) since the subjective evaluation of arousal and his psychophysiological measure may be dissociated52.
Thus, in a random (non-predictive context), the absence of temporal structure appears to favour emotional neutrality, providing an advantage in terms of faster reaction times compared to positive emotional states. This might be because positive emotions, like joy, induce high-arousal processing of stimuli, increasing cognitive load. Conversely, in predictive contexts, arousal-driven mechanisms do not impair motor performance; on the contrary, emotions enhance temporal prediction. Indeed, RTs for joy and fear in predictive contexts (rhythmic and single-interval) were similarly reduced with respect to RTs for joy and fear in the non-predictive context. This suggests that emotions could support predictive anticipation of the temporal features of the stimuli, enhancing the ability to predict future events and to respond more efficiently to anticipated stimuli overshadowing the effect of arousal highlighted only in no predictive contexts. In the rhythmic context, where the predictive advantage is maximized due to the regular structure, all emotional stimuli are equally effective. Conversely, in single-interval contexts, where attentional and mnemonic demands are higher, the benefit is particularly evident for fear, which benefits from its strong role in mnemonic encoding within memory-based contexts, further amplifying its predictive utility.
The effect of biological stimuli on temporal prediction ability
To address differences in temporal prediction due to the complexity of biological stimuli (i.e., faces), we compared RT obtained in the classic TP task employing simple geometric shapes (i.e., squares), with RT obtained in our modified TP emotional task, only with neutral faces.
We found that in the classic TP task RTs were significantly reduced compared to the TP-E task. Such enhancement in RT could be explained by the perceptual complexity of FACS stimuli used in TP-E compared to the colored squares used in the TP task. Previous studies on perception have underlined the hierarchical structure of the visual system. Specifically, lower levels of visual processing are specialized in detecting simple features, while higher visual areas are engaged in processing complex stimuli53. The involvement of later stages of visual processing in the processing of complex stimuli, compared to simple ones, has a distinct impact on behavioral performance54. This idea has been explained by the “stimulus-complexity effect”55 on reaction times, whereby reaction times increase as the complexity of the stimulus increases.
Additionally, a further distinction could be attributed to the difference in the targets used in TP and TP-E tasks. In the TP task the target was a green square compared to a stream of red squares; in this scenario, changing the color could make it easier to distinguish the target from the other stimuli, and fasten motor readiness and response. In the TP-E the targets used were the same stimuli with the addition of a detail, the glasses. This could make it harder to identify the target, by elaborating on subtle characteristics of the target.
It can be assumed that the perceptual complexity of the stimuli and the targets used can have a significant impact on RT due to its influence on the demand for cognitive resources and the perceptual system’s ability to process complex features of the face.
This result seems to be at odds with a strand of literature showing that faces yield privileged processing in cortical and subcortical neural structures compared to other types of visual stimuli (for a review see56). Specifically, electroencephalographic studies have pointed out that faces with an emotional expression elicited an earlier neural response in a pre-attentional and automatic manner57,58. In visual processing, studies have underlined that emotion enhances perceptual processing per se59,60. Neurophysiological evidence reported that emotional stimuli induced an early freezing-like response in the motor system and later motor facilitation29,61,62.
However, it is noteworthy to underline that in the above-mentioned studies, FACS were compared to not embodied stimuli, as in our study. Moreover, at difference with our study, these stimuli were characterized by similar perceptual complexity (FACS vs. emotional scenes taken from IAPS). Here we used stimuli with simple shape (i.e., square), very easy to be process.
Limitation of the study
In this paragraph, we will discuss some limitations of the present study. First, we focused our experiment on two types of emotional FACS (i.e., joy and fear). It would be of interest in future studies to expand this research to other emotional states as anger or shame, or to another set of stimuli, embodied (bodily expression) or not embodied (emotional scenes, IAPS), and to systematically address valence and arousal effect on temporal prediction. Indeed, when the effect of emotion on time judgment was evaluated, it has been emphasized the importance of investigating in detail the dimensions of emotion, both arousal and valence, because the effects of emotion might change in a systematic fashion with dimension changes.
Second, we speculate that the cerebellum may be involved in mediating the decrease of RTs in interval-based predictions, when faces expressing fear were used as stimuli, given the quick involvement of the cerebellum in emotional processing, already at the arousal level and the specific role of the cerebellum in interval-based predictions. However, at this point, this remains a pure speculation, and to address the neuroanatomical network involved in temporal prediction in an emotional context is a topic to be addressed in future studies.
Third, to study temporal prediction in different temporal (single-based and rhythmic) and emotional (fear, happiness, etc.) contexts in patients with basal ganglia or cerebellar disorders would elucidate the role of these subcortical structures in modulating temporal prediction ability, also in relation to emotional stimuli.
