The extent to which biological motion perception relies upon processing of form by the ventral visual system is under debate (e.g., Blake & Shiffrar, 2007; Giese & Poggio, 2003
). Using point-light displays conveying biological and non-biological form from motion, we investigated the ability of LG, a rare case of developmental visual-integrative agnosia with visual integration deficits (Gilaie-Dotan et al., 2009
) to process and perceive biological and non-biological form from motion.
We found LG had normal perception of biological motion, even when tested with point-light stimuli, where the percept of a moving human figure emerges from spatially disconnected local motions of the point-lights, and integration of these into a coherent form is needed. In contrast, he was significantly deficient in processing non-biological form from motion. This pattern indicates that normal biological motion processing can be achieved independently from non-biological form from motion processing. Moreover, it emphasizes the necessity of proper ventral stream function for processing non-biological form from motion.
While we cannot completely rule out the possibility that LG has a subtle deficit in biological motion perception, we believe this possibility to be unlikely, because the paradigms we used here were sensitive enough to detect deficits in biological motion processing in other populations (e.g., in stroke patients, a notoriously heterogeneous sample (Saygin, 2007
)). Moreover, LG demonstrated normal ability to process biological motion in two different experiments featuring different tasks (detection for Experiment 1 and direction determination for Experiment 2), plus in an additional variant of the detection task with a paradigm similar to Experiment 2 (data not shown). Since the functioning of LG's ventral visual stream is deficient, it stands to reason that his normal perception of biological motion relies on his normally functioning dorsal system. Consistent with this, we found normal activation and connectivity patterns in LG's motion sensitive lateral temporal area V5/MT+. LG might also be able to rely on higher brain areas that are part of the APS, such as the STS and premotor cortex (Grossman & Blake, 2001; Puce & Perrett, 2003; Saygin, 2007; Saygin et al., 2004
), as functional connectivity in this network appeared normal in LG's brain. Thus, inputs to the APS from V5/MT+ can be sufficient for normal biological motion perception despite abnormal ventral stream function.
In contrast to his ability to perform well on biological motion tasks, LG showed impairments in processing non-biological form from motion. In Experiments 2 and 3, we found significant differences between LG and controls for non-biological object motion processing. These experiments also utilized different tasks in order to allow us to ascertain any deficits were not task-specific. Whereas the stimuli in Experiment 2 were two dimensional shapes that translated, LG still exhibited difficulty with non-biological object motion when we used three dimensional objects that carried out more object-characterizing movements (Singer & Sheinberg, 2008
). Thus, LG's previously established deficits in form integration also extend to non-biological form from motion perception.
The finding that biological motion processing can dissociate from form integration does not imply that biological motion operates independently of form processing in the healthy brain. In fact, several studies suggest this is unlikely (Lange, Georg, & Lappe, 2006; Lange & Lappe, 2006; Vangeneugden et al., 2009
). However, the present findings show that biological motion can be processed successfully even with compromised ventral stream integration. Perhaps specific for biological motion, the brain appears to be able to compensate for the absence of the normal contribution ventral system makes in perceiving form from motion. It is possible that the visual system may compute biological motion largely relying on a form-based template matching strategy (Lange et al., 2006; Lange & Lappe, 2006, 2007
). In this framework, our data would indicate that these computations can be performed without reliance on ventral stream integration. Vangeneugden et al. (2009)
recently discovered neurons in the STS that appear to be sensitive primarily to body posture rather than to the motion of biological motion stimuli (see also Jellema & Perrett, 2003; Oram & Perrett, 1996
). Given LG's functional neuroanatomy, it is possible that the form processing resources in lateral temporal cortex can be sufficient for biological motion processing. More generally, biological and non-biological form from motion processing may rely differentially on templates that are computed or stored in distinct brain areas (e.g., lateral temporal vs. ventral temporal areas).
Taken together, these data indicate that normal inputs from V5/MT+ can be sufficient for the APS to process biological motion. In high-order ventral cortex, form inputs arriving from retinotopic regions are supported by motion cues from V5/MT+, to create a coherent percept (Felleman & Van Essen, 1991; Grill-Spector, Kushnir, Edelman, Itzchak, & Malach, 1998; Singer & Sheinberg, 2010; Ungerleider & Desimone, 1986
) but LG's case suggests that input from V5/MT+ cannot completely overcome the lack of inputs from intermediate retinotopic cortex.
In conclusion, the present data demonstrate that although form from motion perception from point-light displays requires form integration, it is possible to process biological form from motion even if ventral stream integration is deficient. Our findings extend prior work showing biological motion can dissociate from other kinds of motion perception (e.g., McLeod, Dittrich, Driver, Perrett, & Zihl, 1996; Saygin, 2007; Vaina, Lemay, Bienfang, Choi, & Nakayama, 1990
). In addition, we show that it can also dissociate from form integration. It is therefore possible that there are multiple (and flexible) substrates for biological motion processing, possibly due to the evolutionary importance of the domain.