Inversion Effect For Non-Face Objects

There are two types of cognitive processes in human object recognition: configural processing and feature detection (or local property detection). Are different types of objects recognized by the same process? What factors played decisive roles in the selection of the type of process? All these questions are the focuses of research in the field. In the past several decades, a larger number of studies in this field have used a basic paradigm: an upside-down paradigm. By comparing the recognition performance of the stimulus presented in upright and upside-down orientation, researchers make inference on the types of process employed when they recognized these objects. When the recognition performance is worse when stimulus is presented in the upside-down orientation than upright orientation, an inversion effect is shown. Inversion effect was first discovered in face recognition. Many studies have demonstrated face presented upside-down was more difficult to be recognized by participants than upright face.Using different types of objects as stimuli, the previous researches found that inversion effect only appeared in face recognition, not in common objects. The results indicate a face-specific processing and the configural processing specified in face recognition. However, in recent years, a series of new evidence questioned the face-specific processing theory, because inversion effect also emerged in the recognition of some common objects. For instance, inversion effect was shown when ordinary observers identified static picture of body posture and point-light body movements, when dog experts identified pictures of dog, and when trained participants identified Greeble objects or picture of houses. These results suggested that inversion effect wasn't specific to face and configural processing also existed in common object recognition. Therefore, Reed and her colleagues proposed a configural processing continuum theory to interpret large number of studies on object recognition. But, this theory lacks strong supporting evidence. In particular, there is no theory which can explain why inversion effect showed in some kinds of object recognition, not in others.In the current project, in order to further investigate the mechanism of inversion effect, we used three types of objects: body posture, hand posture and man-made novel object-Dongle. These three types of objects are highly similar in physical properties. Using these objects we manipulated the difference in the degree of experience between upright and inverted exposure of these three kinds of object: the difference of exposure between upright body posture and inverted body posture > the difference of exposure between upright hand posture and inverted hand posture > the difference of exposure between upright man-made novel Dongle object and inverted man-made novel Dongle object. Specifically, the posture of hand is very unique, compared to the posture of body and Dongles. In daily life, average participants have much more opportunity to experience both upright hand posture and inverted hand posture, so the difference of exposure between upright hand posture and inverted hand posture is small. For the posture of body, the average participants have more opportunities of experiencing upright body posture, with less opportunity to experience inverted body posture in daily life, so the difference of experience degree between upright and inverted body posture is larger. However, the average participants never have opportunity of experience the man-made novel Dongle object in daily life, so they don't even know which direction of Dongle is upright and which direction of Dongle is inverted, the difference of exposure between upright Dongle and inverted Dongle is zero. Therefore, if there are any different results among these three types of object recognition, we can attribute them to the difference in exposure between upright and inverted orientation among the three kinds of objects. Meanwhile, we may also gain knowledge on which process the participant employed through whether the inversion effect exists when they recognize these three kinds of objects.In the first study, we used posture of bodies as stimuli. In order to investigate whether body posture recognition demonstrates inversion effect, we controlled the presentation repetition times of body posture and use eye-tracking technique. In Expl, the participants were asked to judge whether sequentially presented two body postures were same or different. The result showed that the upright posture of body was recognized more quickly and accurately than inverted posture of body (inversion effect). The inversion effect appears in the result of the first block of test. These results indicate that body inversion effect did not resulted from repeating presentations. The result of fixation interest area analysis showed that 76 % fixations concentrated on upper body position when participants recognized upright body posture and 59.7% fixations concentrating on upper body position when participants recognized inverted body posture. Combined with the results of the initial fixation duration and numbers of fixation, the results suggest that the efficiency of processing the key information declined greatly when participant recognized inverted body posture, but it still showed the same key information area with upright body posture recognition. The eye-tracking data also suggested that it was a quantitative change of process when upright body posture was recognized compared to inverted body posture was recognized, instead of a qualitative change.In experiment 2, we further controlled the stimuli of body posture in order to explore whether the face or head in body posture led to body inversion effect. The result showed the inversion effect still existed when participant recognized the stimuli of using a black ellipse covered the face part of body posture or removing the head part of body posture. It suggests that it was the body posture itself that led to body inversion effect, instead of face or head. In experiment 3, in order to investigate whether body inversion effect exist in different task, we adopted 10-forced-choice-l task, which indentified more dependent on exemplar level by participants The results showed larger inversion effect in this task compare to the earlier experiment using same-different task. It suggests participants were more dependent on configural processing in this task when they recognized upright body posture. In Experiment 4, we systematically manipulated the degree of deviation from canonical view of the body posture (upright body posture), and found that the performance for recognizing body postures declined with increasing the deviation from the upright body. It shows there was a linear relationship between performance of recognizing body posture and the degree of experience of different view of body posture in picture plane. It suggests it was a quantitative change of processing when upright body posture was recognized compared to inverted body posture was recognized, instead of a qualitative change.In the second study, we used posture of hands as stimuli. In Experiment 5, in order to explore whether hand posture recognition showed inversion effect, we controlled the times of presentation of hand posture and use eye-tracking technique, and the participants were asked to judge whether sequential presentations of two hand postures were same or different. From block 1 to block3, the result did not show inversion effect in reaction time. In contrast, the biomechamcally possible hand postures in block 4 and block 5 and the biomechanical impossible hand postures in block 5 showed inversion effect in reaction time. Eye-tracking data did not show inversion effect. Participant's processing efficiency did not decline when they recognized inverted hand posture. The fixation interest area mainly centered on upper parts of hand for both upright hand posture and inverted hand posture recognition. The initial fixation duration and number of fixation did not show inversion effect. The results of reaction time of experiment 5 ranged from no robust inversion effect to significant inversion effect, which suggested a range of performance from initial mainly depending on feature detection to eventually mainly depending on configural process. In Experiment 6, we adopted repeated tests to investigate whether hand inversion effect stabilized or not. The result showed that the inversion effect remained stable in day 2 and day3 of the tests. In Experiment 7, we used 5-forced-choice-1 task to further explore hand posture inversion effect. The result showed larger inversion effect in this task and emerge inversion effect at first block. In experiment 8, we adopted training program, increasing the degree of experience of upright hand or inverted hand. The result showed that hand inversion effect was based on trained direction.In the third study, we used man-made novel meaningless Dongle and Lingle objects as stimuli. In experiment 9, participants were asked to judge whether sequential presenting two Dongles or Lingles were the same or different, and observe whether the results automatically showed inversion effect like body posture recognition and hand posture recognition did. The results of the five-day test did not show any inversion effect, these results suggested that the difference of degree of experience between upright object and inverted object led to inversion effect from the reverse perspective. In Experiment 10, we used training program, participants were tested only in upright Dongle or in inverted Dongle in training-test stage, then were tested in both upright Dongle or inverted Dongle in post-testing stage. The results showed Dongle inversion effect based on trained direction. This result simulated the body inversion effect and indicated that the difference of experience degree between upright object and inverted object lead to inversion effectTo sum up the result of all the three studies, the results demonstrated a continuum from strong inversion effect to little inversion effect when untrained subjects recognized these three types of object (body posture, hand posture and man-made novel Dongle objects). This suggests that untrained subjects chose appropriate process to complete the recognition task depending on their experience received in daily life. Training participants to experience a specific direction of hand posture or Dongle also led to inversion effect, which suggests the processing method is changeable, both configural processing and feature detection can be selected depending on the past experience. The results of eye-tracking data also showed inversion effect, which was consistent with the result of reaction time, and result of interest area suggests it is not a qualitative change, but quantitative change between upright object processing and inverted object processing.This project represents the first study to examine object recognition through manipulation of the difference in the degree of experience between upright object and inverted object. The result indicates the inversion effect is not specific to face recognition, and common object recognition can also lead to inversion effect. The difference of exposure among object viewpoint in daily life led to object inversion effect. Participants choose method of processing based on their experience when they recognize objects. The results of the whole series of studies demonstrate a continuum from configural processing to feature detection. In order to explain all the results, we propose a theory of interaction among a few factors including the degree of exposure for specific object viewpoint, inversion effect, and type of recognition process. We also provide a revision of the present configural processing continuum theory.