Summary
In this paper the previous claims about cross modal object recognition being viewpoint dependent have been negated. It was observed that the recognition accuracy of the participants did not show any substantial change when objects were presented from different viewpoints cross modally but the accuracy was reduced a lot in the within modal case. However the hypothesis given by the authors in the parent work claimed that cross modal object recognition was mediated by a higher-level representation, which could not be proved in this paper.
Experimental Setup
Fig:Visual Setup used in the experiment
Fig:Haptic Setup used in the experiment
Results
This experiment was able to validate the hypothesis that cross-modal object recognition is viewpoint independent. However no substantial evidence was collected in regard to the second hypothesis of cross-modal object recognition being mediated by a higher-level representation. Acceptable correlation (r=0.5387) was observed between spatial imagery score and the cross-modal unrotated case validating the possibility that cross-modal object recognition could be mediated by an abstract spatial representation which is higher-level in nature. But the unexpected correlation coefficient value (r=-0.1161) for the cross-modal rotated case is an error and the possible reasons behind this discrepancy could be the less number of participants who took the questionnaire (six in number), the abstractness of the questionnaire itself, besides the correlation statistic is very sensitive to noise.
Fig 1: Average observation accuracy (%) versus the modality for both rotated and unrotated cases
Fig 2a:Correlation between Spatial imagery scores and average observation accuracy for each participant; four graphs for both the modalities (within, cross) and orientations (R,UR)
Fig 2b: Correlation between Object imagery scores and average observation accuracy for each participant; four graphs for both the modalities (within, cross) and orientations (R,UR)
References
- Lacey S, Peters A, Sathian K (2007): Cross-Modal Object Recognition Is Viewpoint-Independent. PLoS ONE 2(9): e890. doi:10.1371/journal.pone.0000890
- Olessia Blajenkova, Maria Kozhevnikov and Michaela Motes (2006): Object-Spatial Imagery: A New Self-Report Imagery Questionnaire, Appl. Cognit. Psychol. 20: 239–263
- Maximilian Riesenhuber and Tomaso Poggio (1999): Hierarchical models of object recognition in cortex. Nature Neuroscience, 2(11):1019–1025, November 1999