WEEK 4- ACHIEVING OBJECT CONSTANCY FOR PLANE ROTATION Flashcards

Question

how did Jolicoeur, Corballis and Lawson 1998 compare mirror image to naming tasks (argument A)

Answer 1

-used exactly the same stimuli in object naming and mirror image tasks ie used a horse for both rather than using different objects (weren't able to directly compare). -they tried to push people to rotate in one direction by rotating in 2 degree stages from 120 to 130 to start their mental rotation either clockwise or anticlockwise - if people do use mental rotation this should push them to rotate the object one way rather than the other -CW rotation primes fast responses CCW rotation primes slow repsonses - it is clear people are mentally rotating - anti clockwise was 150ms slower -both still slower than doing the task upright - with the naming task there is a mental cost but doesn't matter if picture is clockwise or not- did not matter which way you encouraged people to rotate -conclusion therefore: mirror image tasks use mental rotation but object recognition tasks do not

Answer 2

-task: word-picture verification- try to identify briefly presented, low contrast, immediately masked pictures of familiar objects -people good at naming tasks if you show them for long enough so instead they showed images very briefly and faded so very hard to see image - they then had to type in what image they thought was presented -almost linear slopes for mirror image task (worst at 180) unlike for naming tasks (instead there is a dip for 180 ie better at 180) but Jolicoeur only tested 60 degrees so the shape of these functions only known coarsely pictures must be shown about 50% longer if they are plane-misoriented by 120 compared to upright - again suggested mental rotation is not used

Answer 3

-the direct evidence involves: A. comparing mirror image and naming tasks B. investigating the shape of the orientation function C. dissociating the effects of visual similarity and object rotation - results in all three cases suggest that the visual system uses different processes to compensate for rotations in identification/ recognition tasks compared to mirror image- discrimination tasks -this suggests that mental rotation is used only in mirror image discrimination tasks and not object recognition tasks - leaves us with the question of how plane-rotated objects are recognised when mental rotation is not used

Answer 4

Corballis- he says that once mental rotation is completed people double check because we are not used to seeing the object in this manner- the double checking is what counts as the cost

Answer 5

- he argued that for mirror image discrimination tasks plane rotated views must undergo time consuming transformation (such as mental rotation) this agrees with the research that we have just reviewed - however corballis claimed that most plane rotated views can be recognised eg named just as efficiently as upright views (he suggested perhaps using view-invariant features) - corballis argued that plane-rotated views are only named slower than upright in recognition tasks because people double check that they have the right answer before they repsond and that they may use mental rotation to do this before double checking

Answer 6

performance becomes flatter as object is rotated because you don't need to keep double checking that the object is a dog- (when the same image is presented (repeated) three times to participants) as people realise their checking largely unecessary - he argues that if people do not check there is no cost to recognising plane- misoriented views so his account predicts that plane- rotation costs on naming reduce when stimuli are repeated (Jolicoeur 1985 also produced this result). - he says this should only affect speeded tasks since double checking should only slow their answers for plane-misorientated views it should not affect accuracy

Answer 7

- by briefly showing low contrast, masked pictures of familiar objects in an unspeeded verification task - contrary to the prediction of corballis' account object recognition was still worse for plane rotated views - conclusion: plane rotation costs are not just caused by double checking - in their study they found the unspeeded task still showed a clear cost of plane- misorientation the first time a picture was recognised - also effects of practice for this unspeeded task, just like for speeded naming tasks - 50% longer if the object was rotated- suggests it is not double checking- this account is being dismissed

Answer 8

- store multiple views for each familiar object; only transform stimuli if they are presented at an unfamiliar view -that usually we store upright views of objects- when we see a plane rotated view that is probably the first time we have seen it- we need to transform the image to uprigtht representation but if we keep seeing these different views of the horse then we will start storing these different views then we can directly match to stored template- multiple views plus transformation account provides a good explanation of most of the data we have discussed.. but what transformation is used?

Answer 9

- most theories of how we achieve object constancy for plane rotation only seem to apply in restricted cases: 1. mirror image discrimination may use mental rotation.. but object recognition does not seem to 2. identifying view invariant features can achieve object constancy, but only in special cases. this strategy is most useful for repeatedly recognising small sets of distinctive objects- but this is rare in everyday life. also double checking does not seem to provide a parsimonious account of how plane-misoriented pictures are recognised 3. with practice, multiple views of an object at different rotations are probably stored. storing these multiple templates reduces the costs of plane rotation for familiar views 4. but how are novel views identified if mental rotation is not the transformation used?