We designed an experimental, descriptive, cross-sectional and quantitative study. Participants carried out tasks whilst undergoing fMRI of the brain. Tasks were based on different established paradigms (with different conditions) but were adapted so that subjects could carry them out within the fMRI machine. The objective was to scan brain activity generated during execution of experimental tasks.

Recruitment was of healthy individuals. Inclusion and exclusion criteria are listed in Table 1. Possible participants completed an initial interview, which included sociodemographic information, and underwent detailed clinical assessment. All participants gave informed written consent to the participation. The study protocol was approved by the Research Ethics Committee of the University of Navarra.

The study sample was formed by 14 healthy volunteers (6 women and 8 men; mean age was 36 years, with standard deviation of 9.29 years), right-handed (except one left-handed subject). None of the volunteers had personal or first-degree family history of psychosis, or any other mental disorder (common or severe).

From the original sample (n = 14), a total of 13 participants successfully completed the study. The data from one of the subjects could not be analyzed due to technical problems, and therefore this subject was excluded from the final results of the study.

We designed our own test to assess TD and OD paradigms. In both paradigms, we built a cognitive control component by introducing two levels of difficulty (difficult and easy). Thus, participants were required to do two tasks (TD and OD) at two levels of difficulty: a total of four tasks (TD easy; TD difficult; OD easy; OD difficult).

Studies were performed on a 3 Tesla MRI scanner (Siemens Skyra; RRID:SCR_020530) using a 20-channel head coil. A T2* weighted echo-planar imaging (EPI) sequence was employed to acquire the BOLD fMRI images. Each volume consisted of 46 slices, with a slice thickness of 3 mm and slice gap of 15%. Time interval between two consecutive acquisitions (TR) was 3 s and echo time (TE) was 30 ms. In-plane image resolution was 3.4 × 3.4 mm², field-of-view (FOV) was 220 × 220 mm², and flip angle was 90°. An anatomical image was also acquired using a 3D T1-weighted MPRAGE sequence with imaging parameters: TR = 2,320 ms, TE = 3.68 ms, inversion time (TI) = 1,170 ms, FOV = 208 × 256× 192 mm³, flip angle = 8°, number of slices = 192, and voxel size = 1 mm isotropic.

BOLD images were pre-processed and analyzed using “Statistical Parametric Mapping” software, version SPM12 (RRID:SCR_007037), in MATLAB (RRID:SCR_001622). All EPI volumes were realigned to the mean and co-registered to the anatomical image. The anatomical image was then normalized to the Montreal Neurological Institute (MNI) template and the normalization parameters were applied to the EPI images. In this step, the EPI images were resampled to an isotropic voxel size of 2 mm. A 3D Gaussian smoothing filter of 8 mm full-width at half maximum was applied to the EPI images.

Statistical analysis was performed at two levels. At the first level, individual subject data were entered into a general linear model (GLM) (Friston et al., 1994). The design matrix was composed of 16 task/control regressors, 6 movement parameters as nuisance regressors and a constant regressor. The 16 task/control regressors model the occurrence of an event type. There were 8 event types: 2 paradigms (TD and OD) x 2 difficulty levels (easy and difficult) x 2 conditions (task and control). The regressors were convolved with a hemodynamic response function (canonical HRF). After model estimation, we calculated statistical contrasts in order to compare brain activation patterns.

For each individual, we performed the following t-contrasts with different objectives:

The first level analysis was followed by group inference using the random effects model (Penny and Holmes, 2003). Random-effects group t-maps were generated for each of the first-level contrasts by applying a one-sample t-test for the contrast parameter values of all the subjects at each voxel. The significance level was set at a cluster-corrected p value of 0.05, with a cluster-generating threshold of 0.001. If map-wise analysis at this corrected level of significance failed to identify areas of activation, results were evaluated using an uncorrected p value of 0.001 with a minimum cluster size of 50.

Additionally, conjunction analysis was performed to determine areas of overlap between TD task and OD task (contrasts 1 and 2). This was done in two steps: first, thresholded activation maps were saved for both contrasts (i1 for TD and i2 for OD, respectively). Subsequently the imcalc tool was used in SPM with the following equation: (i1 > 0) + 2*(i2 > 0). The result was a new image with a value of 3 in areas activated by both tasks, a value of 1 in areas activated by TD only and a value of 2 in areas activated by OD only.

Upon examining the TD task contrast (TD task > TD control), we were able to study the activation relative to performing the discriminatory task (TD task) compared to non-discriminatory activity (TD control). We applied a family-wise error (FWE) correction at the cluster level at a significance level of p = 0.05, with a cluster-defining threshold set at p = 0.001, which resulted in a minimum cluster size of 328 voxels (k = 328). We observed significant activation in several frontal cortical regions. Specifically, the right frontal orbital cortex, inferior frontal gyrus, superior frontal gyrus, and paracingulate gyrus displayed pronounced engagement during the TD task. Additionally, other areas such as the insular cortex and the cerebellum exhibited notable activity, suggesting their contribution in task execution (see Figure 4 and Supplementary Table S1).

Meanwhile, the OD task contrast (OD task > OD control) was subjected to FWE cluster correction with a significance threshold of p = 0.05. The cluster-defining threshold was set at p = 0.001, resulting in a minimum cluster size (k) of 402 voxels. This analysis revealed extensive activation across various brain regions. Increased activity was observed in prefrontal areas, including the supplementary motor area (SMA) and the anterior cingulate cortex. In opposition to the TD task contrast, we observed significant activation in subcortical areas, such as the bilateral putamen. Notably, the cerebellum and posterior regions, such as the precuneus, also showed increased activity. Moreover, there was widespread activation in the bilateral temporal cortex (see Figure 5 and Supplementary Table S2).

After analyzing the contrasts for each paradigm, we performed additional analyses to explore the unique and specific aspects of each task. The TD specificity contrast map (TD task > OD task), utilizing a minimum cluster size (k) of 50 voxels, revealed distinctive activation localized in the right lateral occipital cortex. Interestingly, this region is not typically associated with temporal discrimination (see Figure 6 and Supplementary Table S3).

On the other hand, the OD specificity contrast map (OD task > TD task; k = 223 voxels) revealed significant activation across multiple brain regions. The precuneus cortex, temporal areas, pre-supplementary motor area, occipital cortex, and subcortical areas such as the right putamen exhibited prominent activity. Notably, the brain regions recruited by this task were significantly larger than those engaged in the TD task, confirming the differences observed in the separate analyses of each paradigm (see Figure 7 and Supplementary Table S4).

In addition to investigating the specific differences between the tasks, a conjunction analysis was performed between the TD task and OD task contrast maps. This analysis aimed to examine the overlapping activation and identify common areas of engagement in greater detail. The analysis revealed significant activation overlap in several key regions, including the precentral gyrus, juxtapositional lobule cortex, superior frontal gyrus, paracingulate gyrus, bilateral frontal orbital cortex, bilateral insular cortex, and the cerebellum (see Figure 8). This suggests shared neural substrates and potential interactions between these regions during the execution of both tasks.

The component of cognitive effort was examined by analyzing the differences in activation patterns in response to increasing difficulty levels within each paradigm. The TD effort contrast (TD Task Difficult > TD Task Easy), employing a minimum cluster size of 50 voxels, revealed activation in right prefrontal areas, specifically the right frontal pole and right frontal operculum. These findings suggest increased cognitive engagement and resource allocation in these regions during the more challenging TD task conditions (see Figure 9 and Supplementary Table S5).

However, the OD paradigm did not reveal discernible differences between various levels of difficulty. When analyzing the OD effort contrast (OD Task Difficult > OD Task Easy), no statistically significant differences in the activation patterns were observed. This indicates that the OD task may not induce distinct levels of cognitive effort as detected by the fMRI measurements employed in this study.

Finally, the TD specificity in cognitive effort contrast [(TD task difficult > TD task easy) > (OD task difficult > OD task easy)], minimum cluster size (k) of 50 voxels, revealed activation in the left dorsal dentate nucleus of the cerebellum. This finding suggests that the observed activation in this region is specific to the cognitive effort required during the more challenging TD task (see Figure 10 and Supplementary Table S6).

For each of the two paradigms (TD and OD) we decided to include a control task. During this task (passive listening), participants did not have to perform any discriminative activity (passive listening), but only pressed the same button (“1”), regardless of the perceived stimulus (frequent or infrequent). In this way, comparing the discriminative task with the control task, we were able to assess the evoked activity in each subject in a more detailed way (Sadaghiani et al., 2009; Justen and Herbert, 2018; Kamp et al., 2018).

We describe the pattern of brain activation corresponding to the performance of the TD paradigm. The activation patterns observed (Figure 4) are consistent with those previously reported in the literature (Stevens et al., 2007; Wiener et al., 2010; Ortuño et al., 2011; Nani et al., 2019). In a previous meta-analysis, our research team observed the existence of a cortico-cerebellar-thalamic-cortical circuit, related to time estimation (Ortuño et al., 2011). The same brain areas are the focus of attention for other studies researching temporal processing (Wiener et al., 2010; Matthews and Meck, 2016; Petter et al., 2016; Eichenbaum, 2017; Nobre and van Ede, 2017; Coull and Droit-Volet, 2018; Gili et al., 2018; Nani et al., 2019; Lad et al., 2020; Cona et al., 2021). In conjunction, these findings are consistent with previous evidence for the existence of a brain circuit responsible for temporality.

In this study, we included a non-temporal cognitive task based on the OD paradigm. The brain activity we observed during the performance of OD tasks (greater activity in areas such as the pre-SMA, cerebellum, cingulate cortex and bilateral putamen), was consistent with what is reported in the literature (Ortuño et al., 2005; Stevens et al., 2007; Wiener et al., 2010; Menon, 2015).

Given these two contrasts (the TD activity and the OD activity contrasts) and according to the results of the conjunction analysis between them, we can state that the brain activity corresponding to each of these tasks partially coincides, involving the supplementary motor area, primary motor cortex, prefrontal cortex, insula and cerebellum. However, there are some differences in the activation pattern of both paradigms. In the activity related to the TD paradigm (and not in the OD task contrast), we observed greater activity in the right lateral occipital area, in principle, an area not previously described as involved in temporality. In a previous meta-analysis study (Radua et al., 2014) we found a similar finding, although in that study the additional area can be explained by the fact that the task stimuli used there were visual (and not auditory, as they are in this study). In the study, the presence of the additional activation area can plausibly be explained, as follows, in terms of a complication introduced by the test design. While conducting the experiment, both the researchers and participants felt that the TD paradigm was more difficult than the OD paradigm. We suspect that the possible difference in difficulty between tasks (Alexiou et al., 2009) might trigger the recruitment of other brain areas not related to temporality in order to accomplish the TD task successfully. Occipital cortex activation has been described in both auditory and visual oddball tasks (Stevens et al., 2000; Goldman et al., 2009; McDonald et al., 2013; Alústiza et al., 2018; Justen and Herbert, 2018), which would reinforce the idea that it is precisely in tasks that demand greater cognitive effort that network overlap is more evident. Regarding future studies, we believe that it would be convenient to evaluate in a more objective way, not only the performance in execution, but also the difficulty perceived by the participants during the task (Alexiou et al., 2009; Hay et al., 2022).

When we compare the activation pattern for OD with that for TD, we observe that during OD there was greater activation in temporal regions, bilaterally, and in the precuneus. This observation and the differences in general between TD and OD activation patterns may be due to the study design and the ways tasks were presented. In particular, for the TD paradigm tasks, participants had to focus on the duration of the silence between two sounds, while for the OD paradigm tasks, participants had to focus on the frequency of a sound. So, it is to be expected that brain areas involved in auditory processes would be activated in OD paradigm tasks, as we observed.

Despite the limitations of the study, in view of the fact that the differences between brain activation patterns were small, predictable and also explainable by the limitations of the study design, our results support the hypothesis that the network defined under the TD paradigm has areas in common with other cognitive networks and processes such as OD (Alústiza et al., 2016, 2017, 2018; Garcés et al., 2021).

By establishing easy and difficult modes of the two experimental tasks, we aimed to study the effect of cognitive control on activation patterns. Regarding TD, in the TD effort contrast we found that increased difficulty resulted in implication of right prefrontal regions (specifically, the right frontal pole and the right frontal operculum cortex), which is suggestive of involvement of the dorsolateral prefrontal cortex (DLPFC), which has been previously described as part of a multi-demand system related to more complex cognitive functions and tasks (Tregellas et al., 2006; Lewandowska et al., 2010; Koizumi et al., 2018; Chen et al., 2020; Hertrich et al., 2021). These findings are consistent with the hypothesis that brain networks invoked in TD are engaged in tasks that require greater cognitive difficulty (Livesey et al., 2007; Davalos et al., 2011; Niendam et al., 2012; Alústiza et al., 2016, 2017, 2018).

The OD effort contrast did not show statistically significant differences. This could be because the increase of difficulty in OD task (difference between OD difficult and OD easy) is perhaps not demanding enough to activate the regions involved in greater cognitive effort. In addition, the small sample size in this preliminary study makes it difficult to obtain robust results.

Is there any brain region specific to TD, not active during OD, whose role becomes more important with increasing cognitive load? Interestingly, comparison of the contrast map for increase in difficulty in the TD task with the contrast map for increase in difficulty in the OD task, reveals greater activation in cerebellar regions (see Figure 10). Numerous studies have pointed out the relevant role of the cerebellum in temporal processing and its relationship with areas such as the SMA, parietal cortex, thalamus and basal ganglia (Bengtsson et al., 2005; Koziol et al., 2014; Finnerty et al., 2015; Petter et al., 2016; Bareš et al., 2019). Particularly, detection of the involvement of the cerebellum and subcortical areas in temporal processing is clearer when using tasks of sub-second timing conditions, as in the present study (Petter et al., 2016; Nani et al., 2019).

The pattern of cerebellar activity detected has been reported to be related to motor functions and complex cognitive functions, especially those that involve prefrontal areas (Bernard et al., 2014). This result points at the possibility of the TD network not only being involved in time perception processes, but also being relevant and perhaps playing a specific mediating role in more complex cognitive processes (such as executive functions) (Niendam et al., 2012; Radua et al., 2014; Lošák et al., 2016; Bareš et al., 2019; Pinheiro et al., 2021).

Some of the limitations of this study – the modest sample size, in particular – are inherent to a proof-of-concept study (Kendig, 2015). The proposed theoretical framework of the brain circuits involved in TD processes and their implication in other cognitive functions, being derived from a small sample size, must be taken with caution. Nevertheless, we believe that the insights gained in our experimental study are applicable to larger samples and clinical populations for future research oriented to the implementation and validation of our test-tool. Despite these limitations, the present results are consistent with previous evidence from other task-based neuroimaging studies (Wiener et al., 2010; Niendam et al., 2012; Nani et al., 2019; Garcés et al., 2021), which may help promote larger scale studies.

Another limitation of our study is the age range of participants. To increase sample size and generalizability to the population, we included healthy subjects aged 18 to 65 in our study protocol. In the final sample of participants, the age range was between 23 and 49 years old. However, it is important to note that age-related changes in time discrimination can affect both test performance and neuroimaging results in older subjects (Mioni et al., 2021). In future studies examining temporality and other cognitive functions, it would be beneficial to consider the impact of participant age on the findings. A possible solution could be to include younger subjects, or to expand the sample size to study the effect of age in more detail as a possible confounding factor. An alternative approach would be to compare the findings with performance in other domains (such as processing speed), which have been known to be influenced by age (Baudouin et al., 2018).

In this study, the nature of the tests implied certain limitations: some of the differences in the TD and OD task activation patterns can be attributed to the greater auditory component of the latter task; it is possible that TD task was more difficult to execute than the OD task; and it is hard to determine accurately the increased cognitive load between the two levels of the OD task. We believe, however, that such limitations have not been entirely detrimental but also provided us with valuable information on brain functioning.

The general heterogeneity in neuroimaging research is a big concern. Although this heterogeneity is partly attributable to the range of experimental task employed, a priority in our opinion is to describe differences between individual subjects. To do this, there is a need for further fMRI studies involving more participants and participants with different neuropsychiatric conditions.

Despite recent improvements in the field of neuroimaging, each of the available imaging techniques provides us with a partial view of the real brain structure and function. Integrating different techniques and associating them with fMRI studies can be valuable to study primary domains such as TD. Having looked at the relationship between time perception and change detection, it would be interesting to study the evoked potentials, such as, mismatch negativity (MMN) (Moore et al., 2021). The combinations of findings from task-based fMRI studies and studies that focus on brain structure (i.e., diffusion techniques) may help us to better understand brain connectivity and functioning (de Schotten and Forkel, 2022) especially in disorders such as schizophrenia where aberrant temporal processing has been described (Alústiza et al., 2016, 2017, 2018; Ciullo et al., 2016) and may be related to abnormal structural brain connectivity (white matter impairment) (Ortuño et al., 2011; Gómez et al., 2014; Kochunov et al., 2017, 2022).