Specific disruption of the ventral anterior temporo-frontal network reveals key implications for language comprehension and cognition

Classic models posit that in the left language-dominant hemisphere, language is mediated by a dorsal posterior temporal region (Wernicke’s area) connected to a posterior inferior frontal gyrus region (Broca’s area). The anterior temporal cortex was traditionally not emphasized in language processing. Recent work proposes that the left anterior temporal lobe (ATL) is part of a ventral language network, but its exact role remains debated. The authors investigate whether selective disruption of the anterior lateral temporal cortex and its monosynaptic connection to pars triangularis (area 45) via the temporo-frontal extreme capsule fasciculus (TFexcF) produces a distinct language/cognitive profile compared to lesions of the classic dorsal posterior temporo-parietal system. The study’s purpose is to clarify the ventral network’s contribution—particularly area 45 and ATL—to lexico-semantic processing and controlled retrieval, contrasting it with dorsal system roles in phonology, repetition, and sentence comprehension.

Prior studies have variably implicated the ATL in semantic processing: neurodegenerative conditions (primary progressive aphasia, semantic dementia) often show ATL involvement; unilateral left ATL lesion studies have been inconsistent, with some not confirming semantic deficits, though lexical retrieval deficits and semantic error production have been linked to anterior MTG damage. Anatomically, anterior superior and middle temporal gyri connect monosynaptically with lateral frontal cortex (notably IFG pars triangularis, area 45) via TFexcF, distinct from the uncinate fasciculus and the multi-component inferior fronto-occipital fasciculus (IFOF). The dual-stream framework posits dorsal circuits (AF, SLF II/III) for phonological/sublexical processing and repetition, and ventral circuits (including ATL-TFexcF to IFG) for lexico-semantic processing. Electrical stimulation studies show area 44 (pars opercularis) causes speech arrest, whereas area 45 produces mild naming/word-finding disturbances, suggesting different functions. Functional neuroimaging implicates area 45 in top-down controlled selective retrieval of verbal information. However, selective lesions sparing area 44 while affecting area 45 had not been documented prior to this report.

Design: Single-case comparative study of three right-handed male chronic LH MCA stroke patients: one with a lesion restricted to the ventral anterior temporal network (AA) and two with lesions restricted to the dorsal posterior temporo-parietal language system (MM, TA). A control group of 10 right-handed healthy male adults was included for DTI comparisons. Participants: AA (42M; 1.5 years post-stroke) with anterior STG/MTG lesion up to sulcus acousticus and IFG pars triangularis (area 45) involvement; pars opercularis (area 44) spared. MM (72M; 7 months post-stroke) with posterior STG/MTG, posterior insula, and inferior parietal (SMG/ANG) lesion. TA (72M; 1 year post-stroke) with posterior STG edge and inferior parietal (SMG/ANG) extending to parieto-occipital cortex. All free of other neurological/psychiatric conditions aside from aphasia profiles described. Neuropsychological assessment: Greek BDAE-SF (oral expression, comprehension, repetition, reading); speech rate (Stroke Story; Cookie Theft Picture); Boston Naming Test; PPVT-R; Controlled Oral Word Fluency (semantic and phonemic); Trail Making Test A/B; Pyramids and Palm Trees; Taylor Complex Figure (copy/recall); Imitation of Gestures; Pantomime of Tool Use; Digit Span (forward/backward); Corsi Block-Tapping (forward/backward). Raw scores converted to percentiles per published norms; additional judgments by experienced neuropsychologist where subscale percentiles unavailable. MRI/DTI acquisition: 3T Philips Achieva TX, 8-channel head coil. T1-weighted 1 mm isotropic (TR 9.9 ms, TE 3.7 ms, flip 7°, voxel 1×1×1 mm). DTI: single-shot spin-echo EPI, 30 diffusion directions, TR 7299 ms, TE 68 ms, flip 90°, voxel 2×2×2 mm, b=0 and 1000 s/mm2. Quality check by MRI physicist and neuroradiologist. DTI processing/tractography: Brainance DTI Suite deterministic algorithm with motion/eddy-current correction. ROI-based manual reconstruction of LH tracts: TFexcF, AF, SLF II, SLF III, MdLF, ILF; left cingulum bundle (CB) as control tract. Parameters: step size 1; length threshold 200 mm; angle 35° and FA threshold 0.20 for AF/SLF II/III/TFexcF; angle 27° and FA 0.15 for MdLF/ILF/CB. Multiple-ROI protocols from prior literature (TFexcF per Kourtidou et al.; AF/SLF II/III per Barbeau et al.; MdLF per Kalyvas et al.; ILF and CB per Wakana et al.). Streamlines inconsistent with known connectivity excluded; when lesions obscured landmarks, homologous contralateral ROIs used. Reconstructions performed twice by a blinded rater; subset double-rated for reliability; supervised by an expert neuroanatomist and neuroradiologist. Lesion mapping: Manual delineation on native T1 in MRIcron; normalization to MNI space using SPM12 Clinical toolbox and labeled with an MNI cortical morphology atlas. Statistics: Patient LH tract FA values compared to controls (n=10) using two-sided one-sample Wilcoxon Signed Rank tests (alpha 0.01). Reported p-values and effect sizes for deviations from control medians (Table 1).

• Anatomy/DTI:
  • AA (ventral network lesion): Cortical damage confined to anterior STG/MTG (anterior to sulcus acousticus) and IFG pars triangularis (area 45); pars opercularis (area 44) intact; subcortical involvement of TFexcF pathway. Decreased FA in TFexcF (AA L TFexcF FA ≈ 0.326; p=0.005, r=0.886), indicating disruption of ATL–area 45 connectivity. Dorsal tracts (AF, SLF II/III) spared; posterior temporal and inferior parietal cortices intact.
  • MM and TA (dorsal network lesions): Lesions in posterior STG/MTG and inferior parietal (SMG/ANG). TFexcF preserved. DTI showed decreased FA in AF and/or SLF III; SLF II reconstruction failed due to extensive white matter damage (MM, TA). For example, MM L SLF III FA ≈ 0.355 (p=0.003); TA L SLF III FA ≈ 0.384 (p=0.005) and AF FA ≈ 0.413 (p=0.005). ILF and MdLF FA reduced in MM and TA due to posterior temporal involvement.
• Language/cognition profiles (Table 2):
  • AA: Fluent, errorless speech; intact naming (BNT >50th percentile), intact repetition of words/sentences (BDAE WRp 5/5; SRp 2/2), intact sentence/complex material comprehension (BDAE SC 10/10; CM 5/6; ACi 25th percentile), intact reading aloud and reading comprehension. Selective deficits: markedly low PPVT-R (<5th percentile) and severely reduced semantic and phonemic fluency (COWF-S/Ph <5th), consistent with impaired active, selective controlled retrieval of lexico-semantic information. Working memory (Digits F/B 25th), visuospatial memory (Corsi F 11th; B 28th), construction (Taylor CF copy 50th) largely preserved.
  • MM: Phonemic paraphasias and neologisms; reduced information content; impaired repetition (WRp 2/5; SRp 0/2); impaired sentence/complex comprehension (SC 2/10; CM 0/6; ACi <5th), but single word comprehension relatively spared (PPVT-R 10th; BDAE word comp 15.5/16). Verbal working memory impaired (Digits F/B <5th). Reading largely intact. Dorsal stream hallmark deficits present.
  • TA: Fluent but with phonemic paraphasias, occasional syntactic errors, word-finding pauses, multiple corrective attempts; impaired sentence/complex comprehension (SC 8/10; CM 4/6; ACi <5th); intact single-word PPVT-R (8th percentile). Word repetition spared (5/5), sentence repetition reduced (1/2). Auditory and visuospatial working memory impaired; construction and psychomotor speed impaired.
• Dissociation:
  • Ventral lesion (AA) produced selective lexico-semantic retrieval deficits under higher demand (PPVT-R, fluency) with preserved phonology/repetition/sentence comprehension.
  • Dorsal lesions (MM, TA) produced phonological encoding deficits (phonemic paraphasias/neologisms), impaired repetition and sentence comprehension, and reduced auditory short-term/working memory, with relatively preserved single-word semantics.
• Functional inference: Findings support area 45 and its TFexcF connections with anterior lateral temporal cortex as critical for executive top-down controlled selective retrieval of verbal/semantic information, distinct from area 44’s role in speech motor planning and dorsal pathway phonological processing.

The comparative profiles of a ventral ATL–area 45 disconnection (AA) versus dorsal posterior temporo-parietal lesions (MM, TA) clarify the dual-stream organization of language. In AA, preserved speech production, repetition, and sentence comprehension alongside impaired performance on PPVT-R and verbal fluency indicate a deficit in controlled selective retrieval of lexico-semantic information rather than a general semantic breakdown. This aligns with functional accounts of area 45 (pars triangularis) mediating top-down retrieval control and with TFexcF’s role in linking anterior temporal semantic representations to ventrolateral prefrontal cortex. In contrast, MM and TA exhibited hallmark dorsal-stream impairments—phonological encoding errors (phonemic paraphasias, neologisms), reduced sentence comprehension and repetition, and diminished auditory short-term memory—consistent with disruption of AF/SLF II/III pathways linking posterior temporal and inferior parietal regions to area 44 and premotor systems. Thus, the findings directly address the research question by showing that selective damage to the ventral anterior temporo-frontal network yields specific deficits in lexico-semantic controlled retrieval, while dorsal damage compromises phonological processes and sentence-level comprehension dependent on auditory-verbal memory.

This study provides convergent behavioral and tractographic evidence supporting the dual-stream model of language and, critically, identifies the ventral ATL–pars triangularis (area 45) network as necessary for active, controlled retrieval of lexico-semantic information during demanding comprehension and fluency tasks. It also reinforces the dorsal posterior temporo-parietal system’s role in phonological encoding, repetition, sentence comprehension, and auditory short-term/working memory. Notably, this is the first report of a selective lesion affecting area 45 while sparing area 44, thereby functionally dissociating the two components of classic Broca’s area. Future research should employ higher-resolution diffusion acquisitions and advanced correction methods, expand sample sizes, and include longitudinal designs to assess pre-morbid anatomy and post-stroke reorganization/plasticity.

• DTI acquisition used a relatively low number of diffusion directions and b-values, potentially limiting tract reconstruction fidelity.
• More advanced distortion correction could improve accuracy.
• The chronic-phase case series cannot address pre-stroke anatomical variability or dynamic functional reorganization/plasticity.
• Small sample of rare lesion profiles limits generalizability.