Viewpointdependent and viewpointinvariant theories

Theories of object recognition can be categorised as viewpoint-invariant or viewpoint-dependent. According to viewpoint-invariant theories (e.g., Biederman, 1987), ease of object recognition is not affected by the observer's viewpoint. In contrast, viewpoint-dependent theories (e.g., Tarr, 1995; Tarr & Bülthoff, 1995, 1998) assume that changes in viewpoint reduce the speed and/or accuracy of object recognition. According to such theories, "object representations are collections of views that depict the appearance of objects from specific viewpoints" (Tarr & Bülthoff, 1995). Object recognition is easier when the view of an object seen by an observer corresponds to one of the stored views of that object than when it does not.

Several findings support each type of theory. Research by Tarr supporting viewpoint-dependent theories was discussed by Tarr and Bülthoff (1995). Tarr gave participants extensive practice at recognising novel objects from certain specified viewpoints. The objects were then presented from various novel viewpoints. The findings across several studies were very consistent: "Response times and error rates for naming a familiar object in an unfamiliar viewpoint increased with rotation distance between the unfamiliar viewpoint and the nearest familiar viewpoint" (Tarr & Bülthoff, 1995, p. 1500). These findings are as predicted by viewpoint-dependent theories.

Phinney and Siegel (1999) pointed out that viewpoint-invariant theories (e.g., Biederman, 1987) typically assume that object recognition is based on stored three-dimensional representations of objects. In contrast, viewpoint-dependent theories often assume that object recognition involves multiple stored two-dimensional representations. Phinney and Siegel presented their participants with two random-dot stimuli separated by 1 second, and asked them to decide whether the shapes of the two stimuli were the same. Some of the stimuli contained only two-dimensional cues, whereas others contained only three-dimensional cues. The key finding was that object recognition could be supported by two-dimensional or by three-dimensional cues. Of most theoretical importance, the findings indicate that, "there is an internal storage of an object's representations in three dimensions, a tenet [belief] that has been rejected by viewpoint-based theories" (Phinney & Siegel, 1999, p. 725).

There seem to be some circumstances in which viewpoint-invariant mechanisms are used in object recognition, and others in which viewpoint-dependent mechanisms are used. According to Tarr and Bülthoff (1995), viewpoint-invariant mechanisms are more important when the task involves making easy categorical discriminations (e.g., between cars and bicycles). In contrast, viewpoint-dependent mechanisms are more important when the task requires hard within-category discriminations (e.g., between different makes of car). Indeed, Tarr and Bulthoff (1998, pp. 4-5) concluded that, "almost every behavioural study that has reported viewpoint-dependent recognition has also used tasks in which subjects must discriminate between visually-similar objects, not object classes."

Evidence consistent with this general approach was reported by Tarr et al. (1998). They considered recognition of the same 3-D objects under various conditions across nine experiments. Performance was close to viewpoint-invariant when the recognition task was easy (e.g., detailed feedback on each trial), but it was viewpoint-dependent when the task was difficult (e.g., no feedback provided).

Stop Anxiety Attacks

Stop Anxiety Attacks

Here's How You Could End Anxiety and Panic Attacks For Good Prevent Anxiety in Your Golden Years Without Harmful Prescription Drugs. If You Give Me 15 minutes, I Will Show You a Breakthrough That Will Change The Way You Think About Anxiety and Panic Attacks Forever! If you are still suffering because your doctor can't help you, here's some great news...!

Get My Free Ebook


Post a comment