Why is object recog­ni­tion from novel view­points facil­i­tat­ed if not the object rotates, but the observ­er moves around the object? According to the pre­vail­ing opin­ion, “spa­tial updat­ing” of our mental spa­tial rep­re­sen­ta­tion is sup­posed to be the under­ly­ing process. Here, we pro­vide first evi­dence that chal­lenge this notion, in that dynam­ic visual cues alone might be suf­fi­cient or at least con­tribut­ing to improved object recog­ni­tion from novel viewpoints

Normally, people have dif­fi­cul­ties rec­og­niz­ing objects from novel as com­pared to learned views, result­ing in increased reac­tion times and errors. Recent stud­ies showed, how­ev­er, that this “view-depen­den­cy” can be reduced or even com­plete­ly elim­i­nat­ed when novel views result from observer’s move­ments instead of object move­ments. This observ­er move­ment ben­e­fit was pre­vi­ous­ly attrib­uted to extra-reti­nal (phys­i­cal motion) cues. In two exper­i­ments, we demon­strate that dynam­ic visual infor­ma­tion (that would nor­mal­ly accom­pa­ny observer’s move­ments) can pro­vide a sim­i­lar ben­e­fit and thus a poten­tial alter­na­tive expla­na­tion. Participants per­formed sequen­tial match­ing tasks for Shepard-Metzler-like objects pre­sent­ed via head-mount­ed dis­play. As pre­dict­ed by the lit­er­a­ture, object recog­ni­tion per­for­mance improved when view changes (45° or 90°) result­ed from active observ­er move­ments around the object instead of object move­ments. Unexpectedly, how­ev­er, merely pro­vid­ing dynam­ic visual infor­ma­tion depict­ing the view­point change showed an equal ben­e­fit, despite the lack of any extra-reti­nal/­phys­i­cal self-motion cues. Moreover, visu­al­ly sim­u­lat­ed rota­tions of the table and hidden target object (table move­ment con­di­tion) yield­ed sim­i­lar per­for­mance ben­e­fits as sim­u­lat­ed view­point changes (scene move­ment con­di­tion). These find­ings chal­lenge the pre­vail­ing notion that extra-reti­nal (phys­i­cal motion) cues are required for facil­i­tat­ing object recog­ni­tion from novel view­points, and high­light the impor­tance of dynam­ic visual cues, which have pre­vi­ous­ly received little attention.

links

[mediatags]proj–Teramoto-SU[/mediatags]

Publications