User Manual Privacy Policy Disclaimer Contact us
  Advanced SearchBrowse





A correlation study of behavioral and neural decoding performance for roughness discrimination


Bülthoff,  Heinrich H
Department Human Perception, Cognition and Action, Max Planck Institute for Biological Cybernetics, Max Planck Society;
Max Planck Institute for Biological Cybernetics, Max Planck Society;

Fulltext (public)
There are no public fulltexts stored in PuRe
Supplementary Material (public)
There is no public supplementary material available

Kim, J., Bülthoff, H. H., Kim, S.-P., Chung, Y., Han, S., Chung, S.-C., et al. (2014). A correlation study of behavioral and neural decoding performance for roughness discrimination. Poster presented at 20th Annual Meeting of the Organization for Human Brain Mapping (OHBM 2014), Hamburg, Germany.

Cite as: http://hdl.handle.net/21.11116/0000-0001-32A0-1
Introduction: Recently multi-voxel pattern analysis (MVPA) has been introduced in the analysis of functional magnetic resonance imaging (fMRI) and allows us to examine distributed spatial patterns of neural activity in response to various tactile stimuli [1]. Taking advantage of its higher sensitivity [2], MVPA has been employed in a wide range of somatosensory research fields as a complement to the traditional univariate analysis. However, current research on tactile MVPA is mostly focused on delineating neuronal activation patterns in response to the tactile stimuli. Relatively less attention has been devoted towards understanding how neural activation patterns underlie diverse human behavioral outcomes during tactile manipulation tasks. In this study, we aim to investigate how the multi-voxel neural patterns varied with the behavioral discriminative performance in a roughness discrimination task. For this purpose, we search for the brain regions carrying roughness discriminative information using searchlight MVPA [3] and how each region correlates with the human behavioral performance. Methods: Sixteen subjects participated in this study approved by Korea University Institutional Review Board (KU-IRB-11-46-A-0). Anatomical (T1-weighted 3D MPRAGE) and functional images (T2*-weighted gradient EPI, TR = 3,000 ms, voxel size = 2.0×2.0×2.0 mm) were obtained using a Siemens 3T scanner (Magnetom TrioTim). Before the fMRI scanning, all the participants performed the behavioral roughness discrimination task. Five different roughness levels of aluminum-oxide abrasive papers (Sumitomo 3-M), which were validated and employed in the previous study [4], were used. In each trial of the task, the participants explored two randomly presented abrasive papers with the index fingertip of right hand and reported which of them felt rougher. Behavioral discriminative sensitivity was measured as the difference of roughness values between 25th and 75th percentile of a psychometric function. This was referred to the just noticeable difference (JND) [5]. An fMRI scanning consisted of five blocks with twenty trials. Each trial was made up two consecutive periods; an exploration for 6 s followed by a resting for 15 s. Following instructions, the participants explored a presented abrasive paper with their index fingertip of right hand. Brain signals were analyzed using a searchlight MVPA approach [3] and decoding accuracy of each significant cluster was obtained. Finally, we evaluated a correlation between the JND and the decoding accuracy using the Pearson correlation coefficient. Results: A random-effects group analysis revealed that four clusters exhibited statistically significant decoding capabilities to differentiate five distinct roughness levels (p<0.0001 uncorr., cluster size>30). These four clusters were located in the superior portion of the bilateral temporal pole (STP), supplementary motor area (SMA), and contralateral postcentral gyrus (S1). Decoding accuracies for roughness discrimination were significantly exceeded the chance level (=20) for every clusters (SMA: 40.3±4.4; contralateral S1: 38.0±6.7; contralateral STP: 33.6±4.7; ipsilateral STP: 33.1±3.7). Among these clusters, the significant Pearson correlation coefficient was obtained only for SMA (r=-0.547, p<0.05). Conclusions: In this study, we statistically assessed each set of multi-voxel patterns across the whole brain and revealed that bilateral STP, SMA, and contralateral S1 exhibited neural activity patterns specific to the roughness discrimination. Remarkably, decoding performance using SMA activity showed a significant correlation with the behavioral performance. The negative correlation in SMA indicates that individuals with higher decoding accuracy of roughness from SMA also show better performance in a roughness discrimination task. Our findings suggest that the pattern of activity in SMA may be closely related to the ability to discriminate tactile roughness.