Does linguistic ability impact nonlinguistic learning? The neural bases of nonlinguistic learning in aphasia

• ¹ Boston University, United States
• ² Massachusetts Institute of Technology, Department of Brain and Cognitive Sciences, United States
• ³ Massachusetts General Hospital, United States

Categorization is an important human skill that underlies our ability to recognize and instantly assign meaning to novel items. We focus on probabilistic categorization, a form of rule-based categorization based on the probabilistic association of cues and outcomes (Meeter et al., 2008). Probabilistic categories can be learned via implicit strategies supported by the caudate and putamen. Alternatively, learners can implement verbal and explicit rules, recruiting language regions of the brain, the hippocampus and lateral prefrontal cortex (Ashby et al., 1998). Research has demonstrated that patients with aphasia (PWA) show impaired learning of nonlinguistic probabilistic categories, and that they develop suboptimal strategies when approaching such tasks (Vallila-Rohter & Kiran, 2013, in press). In the current study, we use fMRI paradigms to examine the brain systems engaged as PWA and control individuals learn novel nonlinguistic categories. Methods Stimuli are two sets of fictional animals, organized into two categories based on the distribution of their features. One animal is selected as prototype A, and the animal that differs from that prototype by all features becomes prototype B. Animals must share a majority of their features with each prototype to be considered members of that category. Animals are presented one at a time on a screen and participants guess the animals’ category membership, receiving feedback following each trial. Participants complete 96 classification and 96 perceptual-motor baseline trials that require a button press whenever a unique set of animals appears. Participants also complete a language localizer task in which they read sentences and sequences of nonwords, a task designed to functionally locate language regions of the brain (Fedorenko et al., 2010). Functional and structural images are acquired in a block design using a 3T Siemens TimTrio scanner with a 32-channel head coil. We examine contrasts learning > baseline and baseline > learning and use Marsbar to conduct region of interest (ROI) analyses over regions associated with verbal strategies: MFG, IFG, hippocampus, and over regions associated with implicit strategies: caudate and putamen. Results We have collected data from 6 PWA and two controls and anticipate enrolling 10 PWA. Whole brain analyses reveal a set of regions activated for the learning>baseline contrast across all participants that includes bilateral middle frontal gyrus (MFG), right inferior frontal gyrus (IFG), right angular gyrus and bilateral middle temporal gyrus (MTG). ROI analyses over data for two PWA and two controls demonstrate that controls produce positive percent signal change differences in the caudate (implicit system) and negative percent signal change difference in the hippocampus (verbal, explicit system). In contrast, PWA show positive signal change in the hippocampus and negative percent signal change in the caudate. Conclusions PWA and controls engage a similar overall network of regions during probabilistic category learning. ROI analyses suggest, however, that PWA may exhibit a greater reliance on verbally mediated strategies, compared with a greater reliance on implicit strategies in controls. PWA with mild aphasia who have access to language, may be predisposed to utilize that language to learn, even if it is not a productive strategy.