Abstracts

Rationale:
During a verbal conversation, the human brain moves through a series of complex linguistic stages: decoding of speech sounds, semantic comprehension, retrieval of semantically coherent words, and finally overt production of speech outputs. Each process is thought to be supported by a cortical network consisting of local (e.g. within the temporal lobe) and long-range connections bridging between major cortical areas. In this study, we investigated whether human temporal lobe areas engaged during four language functions are integrated with distant extra-temporal cortical regions via direct effective connectivity. We determined whether the strength of the observed direct inter-lobar effective connectivity was related to the underlying language function, age, cerebral hemisphere, and epilepsy-related profiles.
Method:
We analyzed 3,229 nonepileptic intracranial electrode sites from 37 children (age: 5-20 years) with focal epilepsy who underwent extraoperative electrocorticography recording. We estimated the functional role of each electrode site using the timing of high gamma augmentation during an auditory naming task. We determined which co-activated electrode pairs were supported by direct connectivity using the magnitude of cortico-cortical spectral responses (CCSRs) at the early latency period (10-50 ms) following single-pulse electrical stimulation. To predict the strength of CCSR-based inter-lobar effective connectivity using the high gamma amplitude during each of the four language domain time periods, we employed two sets of mixed models for the connection of temporal and extra-temporal lobe sites, one for each direction. Using the early CCSR and diffusion-weighted imaging data, moreover, we created animations visualizing the trajectory of the physical white matter networks which support rapid propagation of neuronal activity induced by stimulation at the language process-related cortices.
Results:
We analyzed 190 stimulus pairs of adjacent electrodes within the temporal lobe and 460 stimulus pairs in the extra-temporal area. Average per patient, out of 202 pairs of electrodes with co-active high gamma between temporal and extra-temporal lobe sites, 90 had significant early CCSRs in at least one direction (e.g. incoming, outgoing, or reciprocal). A mixed model analysis demonstrated that the strength of high gamma co-activation predicted the magnitude of long-range early CCSR connectivity propagated from temporal lobe to extra-temporal sites (Figure 1A). Older age was associated with more robust direct connectivity from extra-temporal lobe to temporal lobe sites (Figure 1B). The effects of language and age on the strength of effective connectivity were independent of epilepsy-related factors. Dynamic tractography demonstrated that language-related direct connectivity was transmitted mainly via the arcuate fasciculus (Figure 2).
Conclusion:
The human brain may have developed direct inter-lobar connectivity networks allowing rapid transfer of neuronal activity from temporal to extra-temporal lobe regions supporting specific stages of language processing. The human brain may continue strengthening connectivity from extra-temporal to temporal language areas even after age five.
Funding:
:NIH grant NS64033