Abstract
Models of speech perception are centered around a hierarchy in which auditory representations in the thalamus propagate to primary auditory cortex, then to the lateral temporal cortex, and finally through dorsal and ventral pathways to sites in the frontal lobe. However, evidence for short latency speech responses and low-level spectrotemporal representations in frontal cortex raises the question of whether speech-evoked activity in frontal cortex strictly reflects downstream processing from lateral temporal cortex or whether there are direct parallel pathways from the thalamus or primary auditory cortex to the frontal lobe that supplement the traditional hierarchical architecture. Here, we used high-density direct cortical recordings, high-resolution diffusion tractography, and hemodynamic functional connectivity to evaluate for evidence of direct parallel inputs to frontal cortex from low-level areas. We found that neural populations in the frontal lobe show speech-evoked responses that are synchronous or occur earlier than responses in the lateral temporal cortex. These short latency frontal lobe neural populations encode spectrotemporal speech content indistinguishable from spectrotemporal encoding patterns observed in the lateral temporal lobe, suggesting parallel auditory speech representations reaching temporal and frontal cortex simultaneously. This is further supported by white matter tractography and functional connectivity patterns that connect the auditory nucleus of the thalamus (medial geniculate body) and the primary auditory cortex to the frontal lobe. Together, these results support the existence of a robust pathway of parallel inputs from low-level auditory areas to frontal lobe targets and illustrate long-range parallel architecture that works alongside the classical hierarchical speech network model.
Introduction: Challenging the Hierarchical Model
For decades, the dominant model of speech perception has been strictly hierarchical: sound hits the ear, moves to the thalamus, then to the primary auditory cortex (PAC), then to the lateral temporal cortex (for acoustic analysis), and finally to the frontal lobe (for higher-order processing). This serial model implies a significant time lag between temporal and frontal lobe activity. However, this paper investigates whether the frontal lobe has its own "direct feed" of raw auditory data—parallel to the temporal lobe—and whether this data represents low-level sound features rather than just high-level meaning.
Methods
The researchers employed a multi-modal approach to map both the timing of brain activity and the physical "wiring" of the network:
- Electrocorticography (ECoG): Direct neural activity was recorded from 17 neurosurgical patients listening to natural speech (TIMIT corpus), providing sub-millisecond timing.
- Encoding Models: Spectrotemporal Receptive Fields (STRFs) modeled responses to frequencies over time, compared against semantic models (FASTTEXT word vectors).
- Connectivity Analysis: High-resolution diffusion tractography (DTI) and resting-state fMRI from the Human Connectome Project (HCP) were used to map structural and functional connections.
Key Results
The study found that neural populations in the frontal lobe—specifically the middle precentral gyrus (mPreCG), ventral sensorimotor cortex (vSMC), and inferior frontal gyrus (IFG)—showed speech-evoked responses synchronous with or earlier than the temporal lobe.
Early Frontal Responses and Spectrotemporal Encoding
Frontal lobe onset latencies were often less than 80ms. For these short-latency sites (<200ms), STRF (sound-based) models significantly outperformed semantic models ($p < 0.01$). This confirms that early frontal responses encode low-level acoustic "fingerprints" of speech, indistinguishable from primary auditory areas.
Anatomical Evidence for Parallel Pathways
Tractography revealed robust white matter pathways connecting the auditory nucleus of the thalamus (medial geniculate body, MGB) and the primary auditory cortex directly to the frontal lobe targets (IFG, mPreCG, and vSMC).
Discussion
These results support a "long-range parallel architecture" that works alongside the classical hierarchical model. This direct "raw feed" likely facilitates rapid top-down predictions or real-time sensorimotor integration. While the exact functional role remains to be fully elucidated, these findings fundamentally revise our understanding of the human speech processing network.
Summary updated