Event Abstract

Back to Event

Neural model of action-selective neurons in STS and area F5

The visual recognition of goal-directed movements is crucial for the understanding of intentions and goals of others as well as for imitation learning. So far, it is largely unknown how visual information about effectors and goal objects of actions is integrated in the brain. Specifically, it is unclear whether a robust recognition of goal-directed actions can be accomplished by purely visual processing or if it requires a reconstruction of the three-dimensional structure of object and effector geometry.

We present a neurophysiologically inspired model for the recognition of goal-directed grasping movements from real video sequences. The model integrates several physiologically plausible mechanisms in order to realize the integration of information about goal objects and the effector and its movement: (1) A hierarchical neural architecture for the recognition of hand and object shapes, which realizes position and scale-invariant recognition by subsequent increase of feature complexity and invariance along the hierarchy based on learned example views [1,2,3]. However, in contrast to standard models for visual object recognition this invariance is incomplete, so that the retinal positions of goal object and effector can be extracted by population codes. (2) Simple recurrent neural circuits for the realization of temporal sequence selectivity [4,5,6]. (3) A novel mechanism combines information about object shape and affordance and about effector (hand) posture and position in an object-centered frame of reference. This mechanism exploits gain fields in order to implement the relevant coordinate transformation [7,8].

The model shows that a robust integration of effector and object information can be accomplished by well-established physiologically plausible principles. Specifically, the proposed model does not contain explicit 3D representations of objects and the effector movement. Instead, it realizes predictions over time based on learned view-dependent representation of the visual input. Our results complement those of existing models of action recognition [9, 10] and motivate a more detailed analysis of the complementary contributions of visual pattern analysis and motor representations on the visual recognition of imitable actions.

Supported by DFG (SFB 550), EC, and Hermann und Lilly Schilling Foundation.

References

1. Riesenhuber, M. and Poggio, T. (1999): Nat. Neurosci. 2, 1019-1025

2. Giese, A.M. and Poggio, T. (2003): Nat. Rev. Neurosci. 4, 179-192

3. Serre, T. et al. (2007): IEEE Pattern Anal. Mach. Int. 29, 411-426

4. Zhang, K. (1996): J. Neurosci. 16, 2112-2126

5. Hopfield, J. and Brody, D. (2000): Proc Natl Acad Sci USA 97, 13919-13924

6. Xie, X. and Giese, M. (2002): Phys Rev E Stat Nonlin Soft Matter Phys 65, 051904

7. Salinas, E. and Abbott, L. (1995): J. Neurosci. 75, 6461-6474

8. Pouget, A. and Sejnowski, T. (1997): J. Cogn. Neurosci. 9, 222-237

9. Oztop, E. et al. (2006): Neural Netw. 19, 254-271

10. Prevete, E. et al. (2008) Brain Res. 1225, 133-45

Conference: Computational and systems neuroscience 2009, Salt Lake City, UT, United States, 26 Feb - 3 Mar, 2009.

Presentation Type: Poster Presentation

Topic: Poster Presentations

Citation: (2009). Neural model of action-selective neurons in STS and area F5. Front. Syst. Neurosci. Conference Abstract: Computational and systems neuroscience 2009. doi: 10.3389/conf.neuro.06.2009.03.136

Copyright: The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers. They are made available through the Frontiers publishing platform as a service to conference organizers and presenters.

The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated.

Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed.

For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions.

Received: 02 Feb 2009; Published Online: 02 Feb 2009.