Tensor Decomposition for fNIRS-Based Purchase Intention Decoding in Neuromarketing

This study leverages tensor decomposition for the analysis of neural signals recorded via functional near-infrared spectroscopy (fNIRS) to explore the neural correlates of purchase intentions for brand-name products. Tensor decomposition, as opposed to the conventional grand averaging approach in fNIRS signal processing, can identify how hemodynamic response patterns unfold concurrently across time, space, and experimental variables. The dataset comprises fNIRS recordings from 45 college students who were exposed to brand-name products and then completed a purchase intention task. Through tensor decomposition, we were able to identify the unique impact of different study conditions on purchase intention in the bilateral prefrontal cortex and bilateral anterior temporal lobe. These novel findings in the context of neuromarketing validated the utility of tensor decomposition over the traditional grand averaging method. The implications of this research extend to enhancing signal processing techniques in neuroimaging, offering new insights for consumer studies, and potentially informing clinical investigations.