Outline
- Intro paragraph — summarise the functional double dissociation found (FEF→spatial, IFJa→semantic); state that this mirrors Bedini & Baldauf (2021) and supports a supramodal prefrontal architecture
- 5.1.1 The FEF as an Auditory-Spatial Controller
- FEF coupling with IPL (PF, PFcm) and MST as auditory-spatial pathway
- Contrast: SPL connections vanish in partial correlation (unlike visual FEF — shift from retinotopic to multimodal spatial reference frame)
- FEF coupling with STV, TPOJ1 as multimodal convergence hubs
- FEF decoupled from all ventral temporal areas (effect size ≈ 0)
- 5.1.2 The IFJa as a Semantic-Auditory Controller
- IFJa coupling with STS regions (STSdp, STSda) as auditory object identity pathway
- IFJa coupling with BA44 and BA45 → Broca’s language network
- IFJa coupling with A4/A5 → early top-down modulation of auditory features
- IFJa decoupled from parietal spatial areas in partial correlation
- Network membership: language network (Ji et al. 2019 via Bedini & Baldauf 2021)
- 5.1.3 The How-Stream: FEF, 55b, and Auditory-Motor Integration
- FEF premotor cluster (55b, SCEF, FOP1, area 43) from 4.2.2
- Relate to Hickok & Poeppel’s “how”-stream: sensorimotor integration for speech
- Dissociation: FEF = spatial/where controller; 55b = auditory-motor/how controller
- Supported by 4.5 seed-specificity: 55b couples with A4/A5, FEF does not
- 5.1.4 A Supramodal Prefrontal Architecture
- The same two hubs (FEF, IFJa) organise top-down control across vision and audition
- General-purpose attention hubs: spatial/non-spatial axis cuts across modalities
- Evolutionary efficiency argument (Bedini & Baldauf 2021)
- Caveat: resting-state coupling is correlational — top-down directionality requires MEG or effective connectivity (De Vries & Baldauf 2021)
5.1 Resting-State Support for a Supramodal Prefrontal Architecture
Our central hypothesis is supported by our RSFC results (Sections 4.1–4.3), where FEF couples preferentially with auditory-spatial and motion-sensitive areas - the ‘where’-stream. Meanwhile, IFJa couples selectively with the ‘what’-stream, with temporal-semantic and language-related areas. Applying single-seed and comparative partial correlation analyses, we observe a dissociation analogous to the spatial vs. non-spatial segregation previously established by Bedini & Baldauf (2021), suggesting FEF and IFJa as prefrontal control hubs not only for the visual but also for the auditory domain.
5.1.1 The FEF as an Auditory-Spatial Controller
The partial correlation results are consistent with the established role as a spatial attention controller and extend into the auditory domain. Rather than projecting to SPL regions typically associated with visual attention (7PC, 7Am, 7AL), the FEF couples selectively with the IPL, specifically PF and PFcm, and with the motion-sensitive area MST. Through this coupling, the FEF forms a clear auditory-spatial pathway with the IPL and MST, while the SPL connections vanish in partial correlation - unlike in visual analyses with FEF Bedini & Baldauf (2021). The FEF coupling with STV and TPOJ1 suggests that these regions function as multimodal convergence hubs (Rolls (2022) - NeuroImage). Alongside this ‘where’-coupling, the FEF also exhibits decoupling from all ventral temporal areas (e.g. TE1a, TA2, TGd).
This pattern aligns with the functional logic described by Rauschecker & Scott (2009) - Nature Neuroscience, who reviewed evidence that the posterodorsal auditory stream links posterior superior temporal cortex and parietal areas for spatial processing, with activation in regions adjacent to MT/MST specifically for auditory motion. Task-based fMRI supports this: Salmi (2009) demonstrated that top-down controlled shifts of auditory spatial attention recruit FEF/PMC alongside SPL and IPS, serving as evidence for FEF’s functional relevance in auditory spatial orienting. Our resting-state data extend this to the prefrontal level, suggesting that the FEF performs control over auditory spatial orienting, which operates within a multisensory framework (Rauschecker & Scott (2009) - Nature Neuroscience).
5.1.2 The IFJa as a Semantic-Auditory Controller
As shown in Section 4.3, the partial correlation pattern of the IFJa reveals a complementary picture to the FEF, coupling selectively with STS regions (STSdp, STSda), Broca’s areas BA44 and BA45, and early auditory association areas A4 and A5, but not exhibiting substantial coupling with parietal spatial areas. In direct contrast to FEF, this pattern is language- and object-identity-focused, and decouples from spatially parietal auditory regions.
This selectivity is consistent with the IFJa’s network and functional characterisation. Based on the resting-state parcellation of Ji et al. (2019, as reviewed in Bedini & Baldauf (2021)), the IFJa is assigned to the language network, placing it within the non-spatial semantic-language-domain. The results suggest that IFJa may perform top-down attention for auditory feature-extraction, object identity and semantic processing. The coupling with the Broca’s areas could also imply a role in the auditory object-related working memory system Bedini & Baldauf (2021). This is further consistent with the finding that top-down prefrontal control over auditory object processing is implemented via anticipatory alpha oscillations (De Vries & Baldauf (2021) - Journal of Neuroscience).
5.1.3 The How-Stream: FEF, 55b, and Auditory-Motor Integration
As shown in Section 4.2.2, the FEF’s strongest partial correlations are not with auditory regions at all, but with premotor-opercular regions 55b, SCEF, FOP1, and 43. This premotor pattern raises the question of whether the FEF performs auditory spatial control directly or is part of a broader network for auditory-motor integration. Hickok & Poeppel 2007 - Nature argued that the dorsal stream primarily serves auditory-motor integration for translating acoustic speech signals into articulatory motor plans. Our data are consistent with this view at the prefrontal level (Sections 4.2.1, 4.2.4): the FEF maintains spatial coupling with the IPL and MST, while its premotor connections via 55b and FOP1 suggest how-stream functions.
Seed-specificity analysis (Section 4.5) further supports this dissociation. When 55b is used as seed in comparison with FEF, it couples directly with early auditory association areas A4 and A5, connections absent from the FEF’s partial profile. This divergence suggests that 55b functions as the auditory-motor relay for early acoustic features (Dureux (2024)), while the FEF operates on a more abstract spatial coordinate system. Therefore, the FEF may serve as the pure spatial controller, supplying abstract coordinates to the 55b-anchored how-stream without engaging in low-level acoustic processing itself.
→ und hier nochmal single seed 55b anschauen
5.1.4 A Supramodal Prefrontal Architecture
To summarise the global connectivity findings, there is converging evidence that FEF and IFJa function as prefrontal attention hubs not only for vision, but also for audition. One could interpret this as a multimodal organisation (Ghazanfar & Schroeder (2006)), implying that these hubs simply perform attention for both modalities through independent, parallel pathways, processing visual and auditory stimuli separately but within the same regions. We argue for a stronger interpretation: FEF and IFJa are supramodal. Spagna et al. (2015) proposed that executive control of attention operates supramodally, coordinating behaviour across modalities through shared neural mechanisms, instead of alerting and orienting processes, which remain modality-specific. Our RSFC results are consistent with this view at the anatomical level: FEF and IFJa do not merely respond to both vision and audition, but implement the same organisational principle: spatial versus non-spatial top-down control, independent of the sensory modality (Bedini & Baldauf (2021)).
The results suggest a domain-general architecture at the level of attentional control. This is consistent with the evolutionary argument that visuospatial attention systems are preserved in primates and later also recruited other sensory domains Bedini & Baldauf (2021). Whether these RSFC coupling patterns reflect true genuine top-down control signals still needs to be explored with directed connectivity methods (De Vries & Baldauf (2021) - Journal of Neuroscience). The present findings represent a first step towards this hypothesis, further implications are discussed in Section 5.6.