T IME OF DAY EFFECTS IN THE TASK SWITCHING NETWORK Definition of the task switching network

For the definition of the task-related neural activations, the pooled control condition blocks were subtracted from the pooled task switching blocks. Only the resulting activations are reported, please refer to table 4.11. for an overview. The coordinates refer to the peak within significantly activated cluster.

TABLE 4.11. The task switching network: cortical and sub-cortical structures showed a relative

BOLD signal enhancement during the task-switching paradigm as compared to the control condition. All regions are located in the left hemisphere (also indicated by the negative values in the x-plane), unless stated otherwise: {r} corresponds to the right hemisphere. T-values are significant by p < .005. The x, y and z coordinates refer to the Tailarach stereotactic space; BA: Brodmann Area

Compared to the control condition, cognitive performance during task-switching trials activated motor areas, parietal regions, subcortical structures as the thalamus, and the cerebellum. Figure 4.12. depicts the task-switching network.

FIGURE 4.12. The task-switching network as obtained amongst the main study participants

(n=11) comprises the pre-motor area (1), superior (2) and inferior (3) parietal structures and the occipital lobe (5) in the left hemisphere; the precuneus (7) in the parietal lobes, the cerebellum (4) and the thalamus (6) showed bilaterally significant activations. Only activations and no deactivations are shown; left sagittal plan x= -31, middle transversal plane z=30, right sagittal plane x= -11.

Examination of the network stability

First, all areas in the task-switching network were examined with regards to systematic modulation as a function of a) internal time, b) time awake and for c) sequence effects. Second, the non-amplitude corrected data is used to estimate the range of oscillation within a given cluster over 24h. Relative cluster activations of the main study group (outlier and amplitude corrected) were fitted with a two harmonic cosine fit. The following part summarizes the results of the bootstrapping analyses.

Internal time

For the left-hemispheric PMA and the right cerebellum, the best fit between the main and the sup-sample data was obtained with an offset between -1h and +1h. The PMA showed a decreasing trend with a slight increase approximately 13h after individual MSFsc (see fig. 4.13.), as shown by the main study fit r= .61, p < .001, n=43, df= 39. Confrontation with the sup-sample data revealed the best fit at an offset (t offset) of -1h, SS= 232.87.

Cerebellar activation in the right hemisphere decreased steadily as a function of MSFsc, r= .37, p< .05, n=44, df= 40. Least sum of squares was obtained for t offset = +0.25h, SS= 218.84.

FIGURE 4.13. Systematic modulation of motor areas by internal time: the two harmonic fits

obtained on the basis of the main study data (amplitude corrected) is plotted as a function time since ind. MSFsc. Both, the PMA in the left hemisphere and the right cerebellum exhibit the most pronounced relative deactivation around 12h after MSFsc, i.e. for an intermediate chronotype with a MSFsc of 4.5 this would correspond to 16:30 o’clock.

Sequence effects

In the left cerebrum, the BOLD signal in both, the inferior parietal lobe and the thalamus showed sequence effects (inferior parietal area: r= .39, p< .01, n=44, df= 40, t offset = -3, SS= 266.27; thalamus: r= .52, p< .001, n=42, df= 38, t offset = -2, SS= 225.98). Additionally, relative activity was enhanced for the first session in the vermis of the cerebellum with decreasing tendency for the following ones, r= .52, p< .001, n=42, df= 38, t offset = -2, SS= 225.98. Figure 4.14. depicts the sequence effect as modelled by the two harmonic cosine fits.

FIGURE 4.14. Sequence effects: the fit on the main study data obtained the best correspondence with the sup-sample data with an offset of 3h. Thus, a sequence effect in the BOLD response (i.e. in this case an enhanced response for the first session) is assumed. Inferior parietal lobe and the thalamus are both located in the left hemisphere.

The thalamus and the vermis exhibited an enhanced BOLD signal early on, approximately 5h after MSFsc, corresponding to the first scanning session. The inferior parietal lobe showed a similar activation pattern with a relative decrease in the signal, yet, the range of the signal change appears smaller (refer to the next chapter range of oscillation for precise amplitude information).

Time awake effects

Time awake significantly modulated the BOLD signal in parietal structures, the thalamic areas and the right cerebellum. Please refer to table 4.15. for a summary of the results. -100 -50 0 50 100 150 3 5 7 9 11 13 15 Time since MSFsc (h) A ct iv a tio ns (d e vi a tio n fro m in d . me a n b e ta ,%

) Inf. parietal lobe

Thalamus Vermis

TABLE 4.15. Time awake modulations: individual time awake systematically influenced neural activity in the parietal lobe, subcortical thalamic structures and the right cerebellum: the reported r is based on the main study day with a sample size of n; the corresponding degrees of freedom (df) for the two harmonic fit are specified in the second column. T offset = time point of the best fit between the main and the sup-sample data, SS= sum of squares of the best fit. * indicate the significance levels of the respective regressions.

Figure 4.16. illustrates in two panels the activation patterns in the parietal lobe structures (panel A) as well as the thalamic and cerebellar modulation (panel B) in the BOLD signal. Parietal activations followed a similar pattern, revealing a double peak throughout the experimental day. Thalamic and cerebellar activations showed a progressive decline.

FIGURE 4.16. All parietal areas with systematic changes in the BOLD signal over the day were

sensitive to time awake (Panel A); activity modulation in the thalamus is consistent across hemispheres and shows a decrease with a plateau from approximately 5 to 8.5 h after wake-up (panel B). Cerebellar activity decreases steadily. {r} denotes the right hemisphere; the graphs are all based on amplitude-corrected data of the main study group and shows the two harmonic fit. Stable areas within the task-switching network

Two harmonic regression analyses of the BOLD signal in the left occipital lobe and the left cerebellum of the main study group resulted in a significant fit for both, time awake and internal time. Yet, the bootstrapping approach with the sup-

sample data set could not confirm systematic variations for either of them. The best fits between the sup-sample and the main study’s fit laid outside the cut-off criteria defined in the methods part.

Range of oscillation

The range of oscillation within 24h predicted the extent of the modulation within the task-specific network. Table 4.17. gives an overview of the range of oscillation for all significantly modulated structures, may they vary as a function of internal time, time awake or of sequence.

TABLE 4.17. Range of oscillation within the network: the two harmonic regression function of the main study group was used to estimate the range of oscillation for all task-related brain areas exhibiting systematic time-of-day effects; amplitude of the respective structures were highly variable within 24h.

In the case of significant time-of-day effects related to more than one factor, i.e. influence of sequence AND the homeostat, both functions and ranges of oscillations are described.