Generalization in Category Learning: The Roles of Representational and Decisional Uncertainty

Abstract

Effective generalization in a multiple-category situation involves both assessing potential membership in individual categories and resolving conflict between categories while implementing a decision bound. We separated generalization from decision bound implementation using an information integration task in which category exemplars varied over two incommensurable feature dimensions. Human subjects first learned to categorize stimuli within limited training regions, and then, during fMRI scanning, they also categorized transfer stimuli from new regions of perceptual space. Transfer stimuli differed both in distance from the training region prototype and distance from the decision bound, allowing us to independently assess neural systems sensitive to each. Across all stimulus regions, categorization was associated with activity in the extrastriate visual cortex, basal ganglia, and the bilateral intraparietal sulcus. Categorizing stimuli near the decision bound was associated with recruitment of the frontoinsular cortex and medial frontal cortex, regions often associated with conflict and which commonly coactivate within the salience network. Generalization was measured in terms of greater distance from the decision bound and greater distance from the category prototype (average training region stimulus). Distance from the decision bound was associated with activity in the superior parietal lobe, lingual gyri, and anterior hippocampus, whereas distance from the prototype was associated with left intraparietal sulcus activity. The results are interpreted as supporting the existence of different uncertainty resolution mechanisms for uncertainty about category membership (representational uncertainty) and uncertainty about decision bound (decisional uncertainty).

Dynamic updating of hippocampal object representations reflects new conceptual knowledge

“…Indeed, the current study extends prior fMRI work implicating HPC processes in classification learning (22,(51)(52)(53) to demonstrate the influence of goal-relevant selective attention in HPC-based conceptual representations. These results add support for the theoretical proposal from the episodic memory literature that the HPC adaptively builds knowledge representations across episodes (11,12).…”

Mapping Dorsal and Ventral Caudate in Older Adults: Method and Validation

“…Specifically, the DLPFC projects to caudate body and dorsal part of caudate head, while lateral orbitofrontal cortex and anterior cingulate cortex were connected with ventromedial part of caudate. Because of their similar cellular infrastructure and structural connectivity, in recent human studies, caudate body, and caudate tail are combined into one anatomical region (Seger and Cincotta, 2005; Bernácer et al, 2007; Robinson et al, 2012; Seger et al, 2015). Applying probabilistic tractography on fronto-striatal connections, Leh et al found that the anterior part of caudate (head) was strongly connected to ventral lateral prefrontal cortex (VLPFC), while dorsal lateral caudate (body/tail) was connected to DLPFC (Leh et al, 2007).…”

Categorical evidence, confidence, and urgency during probabilistic categorization

“…Within subcortical structures, we also found that the dorsal striatum and dopaminergic midbrain also tracked confidence. These regions are also known to be sensitive to confidence (Ding & Gold, 2010; Schwartenbeck et al, 2014), to be recruited during performance of the weather prediction categorization task specifically (Poldrack, Prabhakaran, Seger, & Gabrieli, 1999; Shohamy, Myers, Kalanithi, & Gluck, 2008), and to be sensitive to categorization demands more broadly (Braunlich, Gomez-Lavin, & Seger, 2015; Lopez-Paniagua & Seger, 2011; Seger, Braunlich, Wehe, & Liu, 2015; Seger, Peterson, Cincotta, Lopez-Paniagua, & Anderson, 2010; Waldschmidt & Ashby, 2011). …”