More Than Words: Extra-Sylvian Neuroanatomic Networks Support Indirect Speech Act Comprehension and Discourse in Behavioral Variant Frontotemporal Dementia

Abstract

Indirect speech acts-responding "I forgot to wear my watch today" to someone who asked for the time-are ubiquitous in daily conversation, but are understudied in current neurobiological models of language. To comprehend an indirect speech act like this one, listeners must not only decode the lexical-semantic content of the utterance, but also make a pragmatic, bridging inference. This inference allows listeners to derive the speaker's true, intended meaning-in the above dialog, for example, that the speaker cannot provide the time. In the present work, we address this major gap by asking non-aphasic patients with behavioral variant frontotemporal dementia (bvFTD, n = 21) and brain-damaged controls with amnestic mild cognitive impairment (MCI, n = 17) to judge simple question-answer dialogs of the form: "Do you want some cake for dessert?" "I'm on a very strict diet right now," and relate the results to structural and diffusion MRI. Accuracy and reaction time results demonstrate that subjects with bvFTD, but not MCI, are selectively impaired in indirect relative to direct speech act comprehension, due in part to their social and executive limitations, and performance is related to caregivers' judgment of communication efficacy. MRI imaging associates the observed impairment in bvFTD to cortical thinning not only in traditional language-associated regions, but also in fronto-parietal regions implicated in social and executive cerebral networks. Finally, diffusion tensor imaging analyses implicate white matter tracts in both dorsal and ventral projection streams, including superior longitudinal fasciculus, frontal aslant, and uncinate fasciculus. These results have strong implications for updated neurobiological models of language, and emphasize a core, language-mediated social disorder in patients with bvFTD.

Keywords: behavioral variant frontotemporal dementia; comprehension; discourse; frontal lobe; frontotemporal dementia; inferencing; white matter tractography.

FIGURE 1

Response accuracy: Response accuracy in controls, patients with behavioral variant frontotemporal dementia (bvFTD), and patients with amnestic mild cognitive impairment (MCI) in the experimental (short) conditions. (A) Mean (±SE) accuracy in the direct and indirect conditions. Controls are shown in dark gray (left-most bar), bvFTD patients in medium gray (middle bar), and MCI patients in light gray (right bar). (B) Mean (±SEM) impairment score (indirect – direct) across groups. A more negative impairment score indicates more difficulty with the indirect condition relative to a patient’s individual baseline performance on the direct condition. * indicates significance at p < 0.05, ** indicates significance at p < 0.01, *** indicates significance at p < 0.001.

FIGURE 2

Response latency: Response latency in controls, patients with behavioral variant frontotemporal dementia (bvFTD), and patients with amnestic mild cognitive impairment (MCI). (A) Mean (±SE) reaction time in the direct and indirect conditions. Controls are shown in dark gray (leftmost bar), bvFTD patients in medium gray (middle bar), and MCI patients in light gray (right bar). (B) Mean (±SE) slowing score across groups. A higher slowing score indicates longer reaction times in the indirect condition relative to a patient’s individual baseline performance in the direct condition. * indicates significance at p < 0.05, ** indicates significance at p < 0.01, *** indicates significance at p < 0.001.

FIGURE 3

Structural Neuroimaging Results: (A) Surface renderings depicting regions of significant cortical thinning in behavioral variant frontotemporal dementia (bvFTD) patients relative to age-matched healthy controls. Heat map intensity refers to t-statistic values. (B) Axial slices and z-axis coordinates illustrating regions of significant cortical thinning in bvFTD patients relative to age-matched healthy controls (red and blue regions) and regions of significant cortical thinning associated with indirect comprehension impairment in bvFTD (red areas, only).

FIGURE 4

Network Analyses: (A) Network Key. Surface renderings of the brain showing each of the 4 network ROIs tested for their relationship with indirect speech processing: language network (green), social network (blue), executive network (yellow), and sensorimotor network (red). See text for a description of how each network was defined. (B) Network associations with Indirect comprehension impairment. Graphs plot the relationships between network cortical thickness in behavioral variant frontotemporal dementia (bvFTD) and indirect impairment score for language, executive, social, and sensorimotor networks. Note that the sensorimotor network is included as a negative control network to demonstrate specificity. See bottom right corner of each plot for R²-values.

FIGURE 5

White Matter Imaging Results: (A) Axial slices showing regions of significantly reduced fractional anisotropy (FA) in white matter (WM) of behavioral variant frontotemporal dementia (bvFTD) patients relative to age-matched healthy controls (blue), regions of significantly reduced FA related to indirect impairment (red), and ancillary white matter regions (outside of blue regions of disease) also related to indirect impairment (violet). See key in upper right hand corner.