Development of Lateral Pulvinar Resting State Functional Connectivity and its Role in Attention

Abstract

Objective

The lateral pulvinar nucleus (LPN) has a well-established role in visual attention. Oscillatory activity of the LPN is critical for cortico-cortical communication within and among occipital and temporal visual processing regions. However, the functional development of the LPN and its role in attention deficits is not understood. This study examined the development of thalamic functional connectivity and its relation to attention abilities.

Method

Resting state functional Magnetic Resonance Imaging images from 950 participants (ages 8–21) in the Philadelphia Neurodevelopmental Cohort (PNC) were used to examine age effects. Follow-up General Linear Models were performed to examine brain-behavior effects with Attention Deficit Hyperactivity Disorder (ADHD) symptom ratings and D-prime scores from the Penn Continuous Performance Task, a behavioral measure of selective attention.

Results

LPN functional connectivity with ventral visual stream regions of the occipital and temporal cortices decreased with age, while LPN functional connectivity with the supplementary motor area increased with age.

Weaker LPN connectivity in the inferior parietal lobule, supramarginal gyrus, posterior insula, and inferior frontal gyrus was associated with more ADHD symptoms; stronger pulvinar-cerebellar connectivity was also associated with more ADHD symptoms. Better D-prime scores were associated with greater connectivity between the pulvinar and superior parietal gyrus; better D-prime scores were associated with weaker pulvinar connectivity with striatal, middle temporal gyrus, and medial prefrontal cortex regions.

Conclusion

These findings implicate the LPN in the development of the ventral visual processing stream between late childhood and early adulthood and suggest that LPN connectivity with higher order attention networks is important for attention abilities.

1. Introduction

The pulvinar is the largest of the three major nuclei in the thalamus, making up approximately one-third of the thalamus’ total volume. It is cytoarchitectonically divided into inferior, anterior, medial, and lateral regions, many of which are closely tied with the various visual processing networks within the brain. Although the understanding of the pulvinar’s structural features is thorough, the understanding of its functional properties is still underdeveloped, especially within the human brain. Recent functional Magnetic Resonance Imaging (fMRI) research reveals that the human thalamus’ function goes beyond a simple relay center, as it is traditionally known as, and encompasses perception and cognition¹. The extensive and reciprocal associations of the pulvinar with cortical areas including the frontal, parietal, temporal, occipital, and cingulate cortex hints at its contributions to the thalamus’ more advanced cognitive capabilities.

Previous lesion studies implicate the pulvinar in visual attention ^1,2. Lesions in humans result in spatial neglect ³, delayed reaction times to targets in spatial attention tasks ⁴, lack of fear recognition and emotional processing ⁵, impairment in feature binding in shape and color ⁶, and inhibition in oculomotor integration ⁷. Disabilities arising from unilateral lesions are confined to the contralesional field. Similar findings were demonstrated in animal lesions ⁸.

Other kinds of studies also indicate that the pulvinar plays an important role in visual attention processes. A study by Kastner and colleagues ⁹ examined the pulvinar through a task fMRI study and observed that the pulvinar was activated when subjects attended to the task stimuli but was not activated when subjects did not actively attend to the stimuli, implicating it in selective attention. However, another task fMRI study by Smith ¹⁰ noted that the presence of the stimulus, not actively attended to, still gave a robust pulvinar response, and that attending to the stimuli resulted in only a modest increase in activation: thus, they did not find sufficient evidence that the pulvinar’s regulation of sensory attention was more significant than in other visual areas. Therefore, while the pulvinar nucleus plays a clear role in visual attention, it is still unclear whether it explicitly regulates attention or only mimics its cortical inputs¹¹.

In particular, the ventro-lateral pulvinar, due to its connections with the ventral-stream visual areas, is a special area of interest because of its potential role in spatial-visual attention and sensory processing ^11,12. Both dorsal and ventral visual pulvinar regions are involved in temporal tracking visual features of task-relevant stimuli. However, as opposed to the dorsal pulvinar, the ventral pulvinar responds robustly to low-level visual and spatial features and to scene-specific visual stimuli ¹³. This is in line with the close connectivity of the ventro-lateral pulvinar with the parahippocampal place area as well as medial regions along the ventral-stream visual pathway ¹¹.

In addition to playing a role in selective attention and visual processing, several findings provide evidence that the pulvinar plays a role in sustained attention. First, as is the case for the thalamus more generally, the pulvinar generates oscillatory activity which is necessary for maintaining communication within and between cortical regions ¹⁴. Second, the pulvinar is connected with arousal-related circuits and, therefore, may be functionally influenced by the arousal state ¹⁵. Third, animal studies indicate that the functional state of the pulvinar nucleus influences performance ¹¹. These findings suggest that the pulvinar is involved not just in visual processing but also in sustaining attention to relevant visual features during task performance.

While several studies associate atypical pulvinar function with inattention found in clinical disorders such as schizophrenia ¹⁶ and attention-deficit hyperactivity disorder (ADHD)^2,17, the neurodevelopment of pulvinar function in humans has not been examined. Resting state fMRI (rs-fMRI) studies of the pulvinar are especially sparse, despite studies indicating its promising role in visual attention. Thus, this study set out to explore the pulvinar’s resting state functional connectivity in relation to both development across age and attention abilities.

The Philadelphia Neurodevelopmental Cohort (PNC), a large cross-sectional neurodevelopmental study, provides an ideal opportunity to examine lateral pulvinar nucleus (LPN) development because of its large community sample of youth that underwent clinical, neurocognitive, and neuroimaging assessments ¹⁸. The age range of participants in this study, 8 to 21 years old, allows insight into the LPN’s changing connectivity over a period of crucial human development and growth. The Children’s Hospital of Philadelphia recruited participants with a diverse range of medical conditions from the greater Philadelphia region¹⁸. Only individuals proficient in English and with adequate physical & cognitive abilities to participate in clinical and neurocognitive assessments were included ¹⁸.

Though there is limited research on pulvinar development, considering its vital role in visual processing, analyses are expected to demonstrate developing pulvinar connectivity with regions involved in various visual networks. Follow-up brain behavior analyses are expected to demonstrate atypical pulvinar connectivity with regions previously implicated in ADHD symptoms.

2. Methods

2.1. Participants

Imaging data from 952 participants of the PNC (431 male, mean age 16.01 years old, SD 3.27 years old; 531 female, mean age 15.90, SD 3.31 years old) ranging from 8–21 years old was processed. The population from the PNC was primarily neurotypical and excluded those individuals with major medical conditions that impair brain function, including neurological conditions and endocrine disorders. Further, participants with a history of learning disabilities, seizures, or loss of consciousness were excluded from the current study sample. To ensure scan quality, T1-weighted and functional images were visually-inspected after MNI standard space normalization to ensure good scan quality (i.e. high signal to noise ratio, no ghosting/artifacts) and that segmentation and normalization preprocessing steps were successful. Participants with framewise displacement (FD) > 0.5 millimeters and participants with less than four minutes of post-scrubbing scan length were excluded from processing and analysis.

2.2. Image Acquisition and Processing

PNC neuroimaging was performed on a single Siemens Tim Trio 3 Tesla, Erlangen, Germany, whole body scanner with a 32 channel head coil¹⁸; participants were scanned for a total of six minutes, eighteen seconds¹⁸. The exact sequence parameters for rs-fMRI imaging were reported in by Satterthwaite and colleagues¹⁸.

Image processing was performed with SPM12 called through MATLAB2015b. These steps included slice time correction, motion correction, co-registration, segmentation, and normalization (Fig. 1). Images were normalized onto the Montreal Neurological Institute (MNI) atlas space. A high-pass temporal filter of 0.005 Hz was applied to all voxel time courses and motion time courses. The aCompcor method of nuisance regression¹⁹ was utilized to mitigate against the detrimental effects of scan motion ^20,21; motion time courses and non-neuronal aCompcor nuisance time courses were regressed out of all voxels. Images were then spatially smoothed (6-mm FWHM) and temporally filtered (bandpass 0.01–0.1 Hz). To further account for the effects of scan motion, time points with FD > 0.2 mm were removed (i.e. scrubbed)²¹. Full-brain connectivity maps were generated for each subject with a 4-millimeter seed placed in the ventro-lateral pulvinar nucleus (MNI coordinates: x = −24, y = −32, z = −2).

Figure 1.

rs-fMRI processing scheme

2.3. Data Analyses

Three models were performed examining: 1) changes in the pulvinar’s connectivity across age, 2) connectivity associated with the ADHD ratings, and 3) connectivity associated with the Continuous Performance Task (CPT) D-prime score. The primary General Linear Model examined 2nd-level group age effects and included age and regressors of no interest: this consisted of regressors to control for confounds like sex, motion (fDcorr: a correlation between framewise displacement (FD) and average volume-to-volume signal change (DVARS)), and scrubbed scan length. fDcorr was used rather than FD, since scan motion may reflect trait hyperactivity, which is likely to correlate with ADHD symptoms; on the other hand, fDcorr is more indicative of the amount of motion-induced artifact that occurs within each participant’s scan. Two follow-up models examining attention abilities (using ADHD symptom scores and CPT D-prime scores) included all of the regressors used in the primary model, in addition to the measure of attention.

2.4. Development of functional connectivity

The primary model examined the linear effects of age to identify developing pulvinar circuits. This model examined age, while controlling for sex, fDcorr, and scrubbed scan length in all 952 participants who passed the image quality criteria. The age contrasts were thresholded at a voxel-wise level of p<0.001 and a cluster wise level of p<0.001^22,23. Since full-brain connectivity of the ventro-lateral pulvinar has not been well-characterized in previous studies, overall positive and negative connectivity of the pulvinar seed was also examined to visualize connectivity of this region. In this case, both positive and negative contrasts were thresholded at a voxel-wise level of p<0.1*10⁻¹⁵ and a cluster-wise level of p<0.001.

2.5. Assessment of Inattention Symptomology in the PNC Sample

According to PNC procedure, participants underwent both a psychiatric assessment and a computerized neurocognitive battery¹⁸. This study utilizes an ADHD rating from the psychiatric assessment and a D-prime score from the computerized neurocognitive battery as measures of attention.

For the psychiatric assessment, interviewers followed the Kiddie-SADS Family Study Interview format that screened based on the DSM-IV and also performed ratings on major disorders, including ADHD, measuring the presence, duration, distress, and impairment of symptoms present ^18,24. ADHD is a clinical neurodevelopmental disorder characterized by impulsivity, hyperactivity, and inattention. Using the questions indicated in Table 1., participants were rated on an overall scale of 0 to 9, depending on the frequency and severity of the behavior being asked of ²⁴. 0 indicates no symptoms and 9 indicates the endorsement of all nine symptoms of ADHD.

Table 1.

Questions used to rate ADHD symptomology

Question 1	Did you often have trouble paying attention or keeping your mind on your school, work, chores, or other activities that you were doing?
Question 2	Did you often have problems following instructions and often fail to finish school, work, or other things you meant to get done?
Question 3	Did you often dislike, avoid, or put off school or homework (or any other activity requiring concentration)
Question 4	Did you often lose things you needed for school or projects at home (assignments or books) or make careless mistakes in school work or other activities?
Question 5	Did you often have trouble making plans, doing things that had to be done in a certain kind of order, or that had a lot of different steps?
Question 6	Did you often have people tell you that you did not seem to be listening when they spoke to you or that you were daydreaming?
Question 7	Did you often have difficulty sitting still for more than a few minutes at a time, even after being asked to stay seated, or did you often fidget with your hands or feet or wiggle in your seat or were you “always on the go”?
Question 8	Did you often blurt out answers to other people’s questions before they finished speaking or interrupt people abruptly?
Question 9	Did you often join other people’s conversations or have trouble waiting your turn (e.g., waiting in line, waiting for a teacher to call on you in class)?

The PNC administered a one-hour computerized neurocognitive battery, the Penn CPT, that measured sustained visual attention ²⁵. This task required participants to press the space bar of a keyboard whenever the computer display formed a digit/letter ²⁵.The number of true responses was used to calculate the D-prime score, or the sensitivity index used within signal detection theory and a measure of selective attention ^25,26. More specifically, it is the hit rate (i.e. proportion of correctly detected targets) minus the false alarm (FA) rate (i.e. proportion of nontargets that were incorrectly responded to) standardized for the current PNC sample. The total number of targets per run was 60 so the hit rate (Hr) = # hits / 60. The total number of nontargets per run was 120 so the FA rate (FAr) = # FAs /120. Both the hit rate and the FA rate were standardized by subtracting the mean and dividing by the standard deviation for the full sample. Therefore, the D-prime value for each subject was calculated as follows: D-primei = [(Hri – Hrmean) / HrSD] - [(FAri – FArmean) / FArSD]. D-prime scores ranged from −6.5 to 2.5, with higher scores indicating better attention capabilities and lower scores indicating weaker attention capabilities ²⁵.

In summary, the PNC ADHD symptom ratings assessed inattention along an increasing scale (higher scores indicate more inattention symptomatology), while the CPT D-prime scores assessed inattention along a decreasing scale (lower scores indicated worse attention behavioral performance). Therefore, a higher ADHD score indicated worse performance (i.e. more symptoms), but a higher CPT d-prime score indicated better performance.

2.6. Functional connectivity associated with attention

Follow-up models were analyzed in order to assess functional connectivity in the lateral pulvinar and its relationship to attention. For the models utilizing ADHD symptom ratings, 948 images were used. 4 participants were eliminated because of faulty ADHD ratings that fell outside of the possible range of values (0–9). Additional regressors of no interest included age, sex, fDcorr, and scrubbed scan length. The PNC is a community sample so rate of ADHD is approximately that of the general population. The majority of PNC participants did not endorse a history of ADHD. Given that the many of PNC participants were expected to have little or no ADHD symptomss and that the PNC ADHD ratings were based on a limited number of questions, which may not accurately represent the spectrum of inattentive symptomatology, the contrasts were thresholded at a less stringent voxel-wise level of p<0.05 and a cluster wise level of p<0.05.

A second model examined another indicator of visual attention capability, the CPT D-prime. This analysis was performed with 744 participants as the D-prime scores for 208 subjects were not available. Additional regressors of no interest included sex, age, fDcorr, and scrubbed scan length. Since previous studies have found modest attention effects in neurotypical populations, the contrasts were thresholded at a voxel-wise level of p<0.05 and a cluster wise level of p<0.05.

3. Results

3.1. Development of LPN functional connectivity

Only one area between the paracentral lobule and medial frontal gyrus (Brodmann Area (BA) 4/6) bilaterally demonstrated a positive association with age, indicating that this region increased in connectivity with the pulvinar over increasing age of participants in the cohort (Fig. 2; Table 2). On the other hand, regions along the occipital lobe extending to the temporal lobe, including the parahippocampal gyrus, significantly decreased in connectivity with the pulvinar with increased participant age. Further, subcortical regions such as the striatum, thalamus, as well as parts of the cerebellum, displayed negative associations with age (Fig. 2; Table 2). Though this negative association was seen bilaterally in most regions, areas that demonstrated the greatest T-scores were primarily found on the left side. In addition to having more regions that showed decreasing connectivity with the pulvinar, the T-scores of these regions tended to be greater than those that indicated increasing connectivity with the pulvinar.

Figure 2.

Changes in LPN over age

Regions of resting state LPN functional connectivity change across age overlaid on overall LPN connectivity. The LPN connectivity model is displayed at a voxel-level threshold of p<0.1*10^−15 and cluster-level threshold of p<0.001(corrected). “Age (+)” (yellow colored) indicates areas with increasing connectivity with the seed region as age increases, while “Age (−)” (green colored) indicate areas with decreasing connectivity with the seed region as age increases. Age effects are displayed at a voxel-level and cluster-level threshold of p <0.001 (corrected). T-score scale indicated on color bar.

Table 2.

Regions with significant age and inattention effects.

Region	Side	BA	Cluster Size	MNI Coordinates (Peak)			Cluster-wise p-value	T-score
Increasing LPN Connectivity with Age
Paracentral Lobule/ Medial Frontal Gyrus	L	4/6	288	−4	−26	56	0.000	4.08
	R	6	~	4	−22	56	~	3.89
	L	NA	~	−8	−16	64	~	3.84
Decreasing LPN Connectivity with Age
Parahippocampal Gyrus	L	NA	10090	−20	−34	−6	0.000	7.59
	R	NA	~	24	−30	2	~	7.30
	L	NA	~	−20	−34	10	~	7.10
Cerebellum, Declive	L	NA	568	−10	−76	−28	0.000	4.92
Pyramis	L	NA	~	6	−80	−32	~	4.51
Inferior Semilunar Lobule	L	NA	~	−20	−70	−46	~	4.25
Greater LPN Connectivity with higher ADHD symptom rating
Cerebellum, Dentate	L	NA	3675	−10	−50	−30	0.000	3.74
Tonsil	R	NA	~	14	−48	−50	~	3.70
Uvula	L	NA	~	−30	−74	−32	~	3.61
Weaker LPN Connectivity with higher ADHD symptom rating
Insula / Inferior Frontal Gyrus	R	13/47	2116	38	10	22	0.004	4.02
Central Opercular Cortex	R	NA	~	52	2	6	~	3.49
Supramarginal Gyrus	R	40	~	50	−16	14	~	3.36
Supramarginal Gyrus	L	40	2562	−64	−42	24	0.001	3.83
Supramarginal Gyrus	L	40	~	−60	−42	32	~	3.49
Angular Gyrus	L	39	~	−38	−42	40	~	3.08
Greater LPN Connectivity with higher CPT D-Prime score
Precuneus	L	7	1660	−12	−76	54	0.018	3.80
Supramarginal Gyrus	L	40	~	−42	−52	50	~	3.41
Angular Gyrus	R	39	~	−34	−42	38	~	3.11
Weaker LPN Connectivity with higher CPT D-Prime score
Nucleus Accumbens	R	34	4504	8	6	−12	0.000	5.31
Superior Temporal Sulcus	R	21	~	46	−16	−12	~	4.66
Inferior Temporal Gyrus	R	20	~	50	−22	−16	~	4.24
Superior Frontal Gyrus	R	10	1846	10	66	16	0.001	3.86
	R	9	~	14	52	20	~	3.85
	R	10	~	10	64	24	~	3.72

BA = Brodmann Area; R= Right; L= Left

3.2. Overall Strength of LPN Connectivity

To visualize overall patterns of LPN connectivity, areas of significant positive and negative connectivity with the seed are underlaid beneath the developmental age effects in Fig. 2 and Fig. 3. The thalamus, anterior cingulate cortex (BA 32), and the cingulate gyrus had a positive relationship to the LPN seed. This positive connectivity also extended into visual processing regions, including a broad area of the occipital and temporal cortices. Areas of the inferior temporal gyrus (BA 20), superior temporal gyrus (BA 22), superior and medial frontal gyrus (BA 10), as well as the supplementary motor area had a negative relationship to the LPN seed.

Figure 3.

LPN Connectivity Associated with ADHD Symptom Scores and the Continuous Performance Task D-Prime Score

Top Panel. Regions in which the strength of resting state LPN functional connectivity is associated with ADHD symptom scores, overlaid on overall LPN connectivity. ADHD symptom score effects are displayed at a voxel-level and cluster-level threshold of p<0.05 (corrected). T-score scale indicated on color bar. “ADHD (+)” (yellow colored) indicates voxels in which there is a positive relationship between LPN connectivity strength and ADHD symptoms (i.e. more connectivity indicates more inattention). “ADHD (−)” (green colored) indicates voxels in which there is a negative relationship between LPN connectivity strength and ADHD symptoms (i.e. more connectivity indicates less inattention).

Bottom Panel. Regions in which the strength of resting state LPN functional connectivity is associated with overlaid on overall LPN connectivity. D-prime score effects are displayed at a voxel-level and cluster-level threshold of p<0.05 (corrected). T-score scale indicated on color bar. “D-prime (−)” (yellow colored) indicates voxels in which there is a negative relationship between LPN connectivity strength and d-prime scores (i.e. less connectivity indicates better attention). “D-prime (+)” (green colored) indicates voxels in which there is a positive relationship between LPN connectivity strength and d-prime scores (i.e. more connectivity indicates better attention).

3.3. Functional connectivity associated with attention

Between the two attention scales, there were certain similarities and differences in the associations with functional connectivity of the pulvinar. The positive association with ADHD symptoms revealed increasing connectivity with the cerebellum, specifically the dentate, tonsil, and uvula areas (Fig. 3; Table 2), while the negative association with D-prime rating revealed increasing connectivity with the nucleus accumbens and the superior frontal gyrus (BA 10) (Fig. 3; Table 2). In this regard, the two ratings shared no areas of commonality. However, there were regions in common that decreased in connectivity with increasing inattention symptomatology between the ADHD and CPT d-prime ratings. For the negative association with ADHD symptoms, regions near the insula and inferior frontal gyrus (BA 13/47) and the supramarginal gyrus (BA 40) regions decreased in connectivity (Fig. 3; Table 2); for the positive association D-prime rating, the precuneus (BA 7) region, which included the supramarginal gyrus (BA 40) and angular gyrus (BA 39), decreased in connectivity (Fig. 3; Table 2). Both analyses revealed peaks within the parietal cortex that were spatially very close. In the ADHD analysis, the peak was found at voxel coordinates: −38, −42, 40, while in the D-prime analysis the peak was found at voxel coordinates: −34, −42, 38.

4. Discussion

The ventrolateral pulvinar is a region that is commonly implicated in visual attention. Like the rest of the thalamus, it plays a role in coordinating and guiding oscillatory activity in cortical networks; however, the ventrolateral pulvinar is specifically connected with ventral visual stream pathways and the medial inferior temporal cortex, through which it maintains low-level visual and spatial representations necessary for task performance ¹³. Further, oscillatory thalamic activity, and ventrolateral pulvinar activity in particular, may play a role in facilitating cortical communication by providing ascending arousal input which improves visual attention^15,27.

Despite numerous human and animal studies implicating this region in attention states, there are no studies examining the neurodevelopment of pulvinar connectivity or its relation to attention abilities in children and adolescents. This study is the first to examine the development of rs-fMRI connectivity of the ventrolateral pulvinar. Development was assessed in a large, cross-sectional cohort of 950 youth, 8–21 years of age. Further, to determine how pulvinar function contributes to attention abilities, the relationship between full-brain LPN connectivity and attention was examined using both a clinical indicator of inattention, as well as a behavioral indicator of sustained attention performance.

These analyses revealed developmental decreases in LPN connectivity with ventral visual stream regions, such as occipital and temporal regions, in addition to several subcortical areas including thalamic, striatal, and cerebellar regions. In addition, LPN connectivity with the Supplementary Motor Area, increased with development. In relation to attention abilities, analyses revealed that for both the ADHD symptom scale measure of attention and the behavioral measure of attention, more connectivity between the LPN and higher-order attention networks indicated better attention abilities.

4.1. Functional development of the LPN

Functional connectivity between the LPN and a circumscribed region of the primary motor cortex and the supplementary motor area increased with development. This developmental increase in connectivity between the visual thalamus and motor regions may allow for maturation of visual-motor coordination advancing from childhood into adulthood. These motor areas were previously implicated in a study of visual-motor associative learning in the trials that tested behavioral expressions to previously learned rules ²⁸. Therefore, while the LPN is primarily involved in coordinating visual processing, pulvinar connectivity with regions of the motor cortex suggests that it may also be involved in the integration of motor information. Strengthening of these LPN motor connections with development suggests that this circuit may play a role in the maturation visual-motor associative learning.

While LPN-motor connectivity increased with development, LPN connectivity to ventral visual stream regions, including the occipital and temporal lobes, decreased with development. The ventral stream has been strongly associated with the lateral pulvinar in previous literature due to the importance it has in visual attention functions ¹². The pathway is involved in object recognition and is especially active during younger ages of development when visual exploration is critical for gaining experience of the external world ²⁹. The pulvinar has connections with multiple regions along the visual processing hierarchy from V1-V4 in addition to inferior temporal regions ^13,30. Its oscillatory activity is involved in both coordinating pulvino-cortical connectivity, but also in modulating the synchrony between visual cortical regions ³¹. The current findings suggest that pulvinar connectivity with ventral visual stream regions may be more critical in establishing synchrony with ventral visual stream cortical regions earlier in development.

In addition to these ventral visual stream regions, developmental decreases in LPN connectivity occurred with a number of subcortical regions, including the thalamus, striatum, and arousal brainstem nuclei. Developmental decreases in local subcortical connectivity have also been found for striatal connections ³². This stronger local subcortical connectivity early in development may reflect the early establishment of subcortical connections. It could also be that subcortical involvement in cognitive processing is more critical early in development when higher-level cognitive networks are less established. The role of thalamus in facilitating cortical communication is in line with this interpretation.

Developmental decreases in pulvinar connectivity also occurred within the declive areas of the cerebellum. This cerebellar region is commonly associated with saccadic eye movements³³ suggesting that these connections are important for visuomotor integration. As with pulvinar connectivity to other subcortical regions, its connectivity with the cerebellum was stronger earlier in development suggesting that these connections play a stronger role during childhood than later in development. Although this suggests that pulvinar connectivity to motor regions is more important for the early development of visuomotor learning, the finding that pulvinar connectivity with the supplementary motor area increased in strength with age suggests that there may be a developmental restructuring of visuomotor connections.

4.2. Overall LPN Connectivity

The LPN seed was positively correlated with a broad area of the occipital and temporal lobes, including the multiple regions of the visual cortical hierarchy, in line with the role of this region in modulating and integrating visual information across visual cortical regions ³⁰ and is similar to the findings of previous rs-fMRI studies in humans³⁴ as well as animal studies. One study in particular examining the pulvinar in prosimian primates concluded that the results of manipulating the lateral pulvinar’s input strength can vary from extinguishing a response completely in its absence or strongly amplifying responses in its attendance, making the pulvinar integral to the visual cortex’s function³⁵. These connections may serve the foundation of the pulvinar’s proposed role in bottom up salience in visual attention. Additionally, the pulvinar was significantly functionally connected to the cingulate cortex, most prominently in the anterior region or BA 32, an area not commonly associated with the lateral pulvinar. Interestingly, this region is activated during the Stroop task and thus attention processes involving selective visual attention ³⁶.

While the pulvinar seed was positively correlated with the midbrain stretching to the posterior cortical regions, it was negatively correlated with a few anterior cortical regions. In particular, BA 9/10, or portions of the medial prefrontal cortex (MPFC), were significantly anticorrelated. This MPFC region is part of the Default Mode Network (DMN), which is anticorrelated with higher order attention and cognitive control networks ^37,38 and is involved in internally, self-directed thought ³⁹. Therefore, negative connections between the pulvinar and MPFC may facilitate externally directed visual processing by suppressing self-referential information.

4.3. Increased pulvinar connectivity associated with inattention

To determine whether pulvinar connectivity contributes to attention abilities, associations with pulvinar connectivity were examined for two different attention measures. One was a measure of ADHD symptoms that mainly reflected inattention, and the other, the CPT d-prime score, was a behavioral measure of sustained attention.

Within the positively associated ADHD symptom score analysis, the pulvinar increased connectivity within the cerebellum, especially the dentate nucleus. Previous ADHD fMRI studies have also noted an increase in cerebellar activation in ADHD patients as opposed to controls ⁴⁰. This increased activation within ADHD patients is thought to be a compensatory response in decreased activation in the dorsolateral-PFC in the dorsolateral-PFC-cerebellar networks of sustained attention ^40,41. Rs-fMRI studies have reported that increased activation of cerebellar regions tends to occur alongside temporal and striatal regions ⁴². This aligns with the associated activation patterns of temporal and striatal regions reported in the d-prime analysis in which increased pulvinar connectivity was associated with worsening attention. Furthermore, decreased cerebellar connectivity with development and increased cerebellar connectivity with inattention symptomatology may indicate that those with inattention display relatively immature pulvinar-cerebellum connectivity. In regards to volumetric developmental differences in ADHD, a multicohort, longitudinal study of cerebellar development in those with ADHD noted that cerebellar growth was slow in early childhood (4–8 years old) as compared to late childhood (8–12 years old) ⁴³. This growth period coincides with the PNC’s age range and implicates the pulvinar in developing functional connectivity networks alongside volume-wise change.

Within the negatively associated D-prime analysis, increased pulvinar connectivity with areas along the temporal lobe and the superior frontal gyrus was associated with worse behavioral attention. Considering the role of these regions in the DMN, this pattern of development may be due to mind-wandering when the DMN is not adequately suppressed during visual attention. This is consistent with the view that increased DMN activity interferes with cingulo-fronto-parietal attention networks and noted the DMN as a potential target for ADHD medication^{44, 45,46}. Developmental resting state functional connectivity studies have noted that atypical consolidation of the DMN is an important contributor to ADHD symptoms ⁴⁷. Furthermore, cortical volume-wise development and maturation of these areas (in addition to the cerebellum) within children ADHD is delayed in comparison to control peers ⁴⁸. This latter development may result in increased connectivity during the point of new development as the brain begins to establish more connections. The superior frontal gyrus has also been noted for its distinct and significant volumetric differences between ADHD patients and controls. More specifically, the region’s volume is significantly smaller in ADHD patients ⁴⁹, which may necessitate a stronger positive association between the pulvinar and the region to compensate for the smaller volume. Other resting-state functional connectivity studies have similarly implicated a positive association with the superior frontal gyrus in ADHD ⁵⁰.

4.4. Decreased pulvinar connectivity associated with inattention

While the two measures of attention used in this study reflected different aspects of attention, and generally were associated with different pulvinar connections, some interesting similarities existed among the associations with the two measures of attention. Both the ADHD symptom ratings and the CPT d-prime score were associated with pulvinar connectivity to higher-order attention networks such that weaker pulvinar-attention network connectivity reflected worsening attention abilities. For the ADHD symptom rating, the inferior frontal gyrus, insula and inferior parietal lobule were the primary areas of peak activation. The inferior frontal gyrus is part of the ventral attention network that has been commonly implicated in ADHD inattention symptomatology both in regards to volume ^51,52, and in functional connectivity ^50,52. It also has a role in motor response inhibition, suggesting that a reduction in activation in this region contributes to more than one ADHD symptom ⁵³. In a study that also used the PNC sample, development along the striatal-inferior frontal gyrus networks were strong indicators of ADHD symptoms ³². Intermediary regions between the insula and frontal gyrus is also considered a key region in the ventral attention network, and may play a role in attention-related problem solving ⁵⁴. Further, other regions of the ventral attention network that also displayed developmental decreases in activation, the supramarginal gyrus and the angular gyrus, were also associated with inattention in the follow-up d-prime analysis. However, for the associations with the behavioral d-prime score, this parietal region extended more dorsally and the peak was in the precuneus, which is more consistent with the dorsal attention network ⁵⁵. Therefore, these two distinct measures of attention abilities were both associated with connectivity between the pulvinar and cortical attention network regions.

Previous studies have found that decreased activation in these parietal areas, and more broadly in the inferior parietal lobule, are often associated with clinical functioning deficits in time discrimination, as well as selective, divided, and sustained attention ⁵³. Similarly, a multicohort resting state functional connectivity study has noted a negative association between connectivity between the dorsal attention network and the supramarginal gyrus, angular gyrus, and precuneus among children with ADHD ⁵⁰. A task-based fMRI study, in which the inferior frontal gyrus was shown to have stronger associations with the pulvinar in normal controls as compared to ADHD patients, confirms that pulvinar and higher-order attention function is atypical in children with ADHD¹⁷. Atypical development of the inferior frontal gyrus is also prevalent in ADHD patients ⁵⁶. A study using rs-fMRI ⁵⁷ noted that the inferior frontal gyrus and inferior parietal cortex becomes more anticorrelated with the DMN with development and these connections are important for inhibitory control abilities in adulthood. This is also consistent with the relationship between d-prime scores and pulvinar-MPFC connectivity in the current study. Those participants that had weaker connectivity between these two regions, had better attention abilities. Since the MPFC forms part of the DMN, and plays a role in self-directed thought, it makes sense that stronger anticorrelation with the pulvinar, a region involved in externally-directed visual attention, would be related to better attention. Further, the finding that the pulvinar-MPFC association with attention was opposing to that of the pulvinar-inferior frontal lobule/precuneus, is consistent with the opposing relationship between externally-directed dorsal attention network regions and internally-directed DMN regions ³⁷. The relationship between lateral pulvinar connectivity with higher-order cognitive networks and attention suggests that these connections play an important role in attention.

5. Conclusions

This is the first study examining the functional neurodevelopment of the ventrolateral pulvinar nuclear of the thalamus, a region that has been previously implicated in visual attention. The use of a large, representative population cohort, sample spanning an important development period, 8 to 21 years old, allowed for the examination of linear effects of age using rigorous statistical approaches with conservative significance thresholds.

However, although the PNC is a great resource, its cross-sectional nature limits the conclusions that can be drawn in regard to development. Furthermore, while this study had an a priori reason to explore this particular region and its role in attention, further studies should take a more comprehensive approach at looking at functional subregions of pulvinar or thalamic development.

In conclusion, this study implicates the pulvinar as a player in developing attention networks across the brain and demonstrates its contributions to higher-order cognitive capabilities.