View dependence in scene recognition after active learning

Abstract

Human spatial encoding of three-dimensional navigable space was studied, using a virtual environment simulation. This allowed subjects to become familiar with a realistic scene by making simulated rotational and translational movements during training. Subsequent tests determined whether subjects could generalize their recognition ability by identifying novel-perspective views and topographic floor plans of the scene. Results from picture recognition tests showed that familiar direction views were most easily recognized, although significant generalization to novel views was observed. Topographic floor plans were also easily identified. In further experiments, novel-view performance diminished when active training was replaced by passive viewing of static images of the scene. However, the ability to make self-initiated movements, as opposed to watching dynamic movie sequences, had no effect on performance. These results suggest that representation of navigable space is view dependent and highlight the importance of spatial-temporal continuity during learning

Geometric determinants of human spatial memory

“…These findings suggest a special role for vestibular or proprioceptive signals in spatial memory consistent with the use of egocentric representations of the objects' locations combined with an automatic updating process driven by internal cues to self-motion. However, some studies indicate that a similar kind of cumulative updating process may occur at the point of retrieval, rather than concurrent with movement (Diwadkar & McNamara, 1997), or even in the absence of physical movement (Christou & Bulthoff, 1999;Wraga, Creem, & Proffitt, 2000). In addition, there is evidence that the egocentric representations of different objects are updated independently, in that the additional error in pointing to object locations due to disorientation varies more between objects than would be expected from the mean withinobject variation (Wang & Spelke, 2000).…”

Spatial Reasoning With External Visualizations: What Matters Is What You See, Not Whether You Interact

“…Finally, in studies on searching for structure in 3D data (Marchak & Zulager, 1992) and tactile maze learning (Richardson, Wuillemin, & MacKintosh, 1981), participants who were given active control were found to perform worse than passive participants. Even within a single study, different comparisons of active and passive conditions have sometimes produced apparently contradictory results (Attree et al, 1996;Christou & Bülthoff, 1999;P. N. Wilson & Peruch, 2002).…”

“…Monitoring motor commands may provide especially strong cues about spatial properties (e.g., Christou & Bülthoff, 1999;Feldman & Acredolo, 1979;Philbeck, Klatzky, Behrmann, Loomis, & Goodridge, 2001;Wang & Simons, 1999), particularly with a naturalistic interface designed such that manipulations made by users are exactly mirrored in the resulting movements of the visualization (cf. Schneiderman, 1983).…”

How Are the Locations of Objects in the Environment Represented in Memory?

“…Participants have learned several views of layouts; have learned layouts visually, tactilely, via navigation, and via desktop virtual reality; have been tested in the same room in which they learned the layout or in a different room; have been oriented or disoriented at the time of testing; have been seated or standing during learning and testing; and have been tested using scene recognition, judgments of relative direction, or both (e.g., Christou & Bülthoff, 1999;Diwadkar & McNamara, 1997;Easton & Sholl, 1995;Levine, Jankovic, & Palij, 1982;Mou & McNamara, 2002;Presson & Montello, 1994;Richardson, Montello, & Hegarty, 1999, map & virtual-walk conditions;Rieser, 1989;Rieser, Guth, & Hill, 1986;Roskos-Ewoldsen et al, 1998;Shelton & McNamara, 1997, 2001a, 2001b, 2001cSholl & Nolin, 1997, Exps. 1, 2, & 5; Simons & Wang, 1998).…”