Assignment Coversheet
Please complete all sections below and attach the completed coversheet to the front of your electronic assignment before submission:
Please complete before submitting your assignment:
First Marker’s comments
Please see comments below for further feedback.
First Marker______________________ Date__________ Mark__________
Moderator_________________________ Date__________ Mark__________
Agreed Mark____________
Abstract
The presence of colour in visual display either surface colour or perceptual plays an important role in assisting recognition. However, this assumption is widely debated because some edge-based theories are arguing that perceptual colour is not particularly helpful whilst other theoretical claims suggests that surface-based is somewhat valuable in providing salient clue. Due to the variation in strop paradigms in which the role of colour indeed is helpful in determining recognition, this study will examine the physical properties of objects and their encoded visual representation through colour. Sets of cognitive processes that insinuate the changes in perceived images analysed by the retina will be used to determine whether the visual stimuli can be recognized as the same familiar object. In order to achieve the initiatives of this study relevant tool and methodology will be employed to establish the important role of colour to cognitive recognition. A set of participants will also be employed in replicating Naor-Raz et al. (2003) experiment by adding variables such as blurring. The results of the experiment will be evaluated using the 2x2 two-way within subjects ANOVA study design. In addition, the experiment will also use computer-run programme called E-prime script. It is a set of coloured pictures from everyday objects obtained from Naor-Raz et al. (2003) original experiment along with other materials form TarrLab. A Total of 80 pictures will be used wherein 20 pictures will be used for various conditions such as typical-clear, atypical-clear, typical-blurred and atypical blurred.
Introduction
Visual recognition of objects is an important ability for all living organisms. It allows categorization of objects in which an organism can behave suitably in different cases such as running away from or reaching for (Yantis, 2001). The ability to recognise objects through vision is an essential part of our everyday lives and it happens unconsciously, effortlessly and rapidly (Logothetis &Sheinberg, 1996). According to Logothetis and Sheinberg (1996), there are many questions that have been addressed concerning the visual recognition of objects such as how does the visual system stem from the retinal image to create a description of sets specific features of objects that sum up all the objects invariant properties. There are sets of cognitive processes in which it changes the object’s basic description that is created by the analysis of the retinal images into familiar objects that can be recognised as specific objects (Pike & Brace, 2012). In their model of object recognition that fits a broader context of cognition, Humphreys and Bruce (1989) suggested that there are series of stages for objects to be recognised (Pike & Brace, 2012). When the sensory input is generated (early visual processing) it leads to the perceptual classification. Perceptual classification is when previously stored description of objects is compared with the new sensory input. Later, when objects are recognised and categorised, it leads to the next stage where items are semantically classified, consequently naming the objects (Pike & Brace, 2012). Computational scientists such as Marr (1982) and cognitive psychologists such as Biederman (1987) try to address this problem and proposed theories for object recognition that try to explain the stages of perceptual and semantic classification (Logothetis &Sheinberg, 1996).
According to Pike and Brice (2012), the foremost problem that faces all the theories of visual recognition is that retinal image is perceived as a 2D image, although objects are 3D. Marr and Nishihara (1978) proposed a theory of object recognition that is based on generating 3D object-centred models which allows object to be recognised from any angle. Marr and Nishihara (1978) argued that generalized cones or cylinders are based on objects it self, “ object-centred”. Generalized cones are “any 3D shape that has a cross-section of a consistent shape throughout its length” (Pike & Brace, 2012, p109). Therefore, if the viewpoint of the observer changes relativity to the object, the description of an object would not be invariant (Marr & Nishihara, 1978). Their object recognition theory was achieved through representing three-dimensional shapes, those include; (a) from a 21/2 D sketch, shapes of an objects were derived, (b) dividing the shapes into primitives based upon areas of concavity (concavity where two line come-in together), (c) for each of these components, an axis is generated, (d) and finally through generalized cones, every components was presented (Marr & Nishihara, 1978). Later, Marr (1982) proposed his theory of computational theory because of he believes that vision “no more than an information-processing task” (Marr, 1982). Marr (1982) proposed four stages from the retinal images to internal object recognition; (1) the retinal image, (2) the primal sketch, where edges, light changes and contours are made unambiguous, (3) a 21/2 D sketch. This includes information about depth, motion and shading, (4) and finally 3-D model representation of objects that us built up hierarchically and it is observer independent (Marr, 1982). Following up to Marr (1982) theory, Biederman (1987) extended the work of Marr and Nishihara (1978) in a similar theory but instead of restricted generalized cones of he introduced a set of 36 common geons “geometric ions” for object recognition and for the spatial relations between them, this is also known as Recognition by Components (RBC). Biederman (1987) proposed that information in a 2D image could be generated to a 3D image. His approach was developed by identifying 5 constrains of object recognition. According to Biederman (1987) an object can be (1) recognized rapidly, (2) can be recognized from different viewpoints, (3) under moderate levels of visual noise such as different lightning conditions, (4) object can be recognised even if partially occluded, (5) and when it is a new exemplar of a category such as different chairs designs (Biederman, 1987). Biederman (1987) suggested that geons are the key features that remain unchangeable across different viewpoints. There are 5 non-accidental properties that were suggested by Biederman (1987) that are assumed to invariant across conditions. These are collinearity, curvilinearity, parallelism, symmetry, and cotermination (Biederman, 1987).
However, because the RBC has different limitation as it does not take into account different questions such as the role of colour. Biederman and Ju (1988) argue that colour, texture or fine details have no role in object recognition. Tanaka and Presnell (1999) argue that in contrast to Biederman and Ju’s (1988) opinion, colour is widely acknowledged as it has an important role of object recognition. According to Tanaka & Presnell (1999), the impact of colour depends on certain conditions. For example how a typical a colour is of an object has an influence of faster recognition of the objects, this is known as colour diagnosticity (Tanaka & Presnell, 1999). Furthermore, Tanaka et al. (2001) argued that “any mental representation of an object is invariably a combination of all this information”, Information includes colour, shape and texture.
Noar-Raz et al. (2003) conducted a study that aimed to investigate the relationship between colour and how objects are mentally represented. His hypothesis, that colour and shape are intrinsically related in mental representation of object was based on the early evidence that was suggested by Biederman and Ju (1988) that colour is not important when recognising objects, and later to support Tanaka and Presnell (1999) that colour is important in certain conditions (Naor-Raz et al, 2003). The task type in his experiment was object-naming task that requires semantic information about objects. Object-Stroop experiment was used to test if colour and shape are mentally represented together. The experiment had 22 undergraduate participants at Brown University. The design of their study was to measure the speed of colour identification by the typicality of object colour. It was found that there was a significant difference between condition, which concluded that colour and shape of objects are intrinsically linked and are not processed independently.
For this reason, a part replication experiment of Naor-Raz et al. (2003) was conducted but with adding another independent variable “blurring”. The aim of this experiment is to support Naor-Raz et al. (2003) that colour have an important role in object recognition and it is expected that there will be a significant interaction between the role of colour in object recognition and if adding another variable “blurriness” will have an effect on it.
Method
Participants
The participants for the current experiment consisted of 20 undergraduate psychology students (7 males, 13 female) aged between 18-23 years old. Participants were recruited using opportunity-sampling technique and they were all Caucasian and English was their first language. Each participant was given a consent form and was informed of his or her right of withdrawal at any time during or after the experiment.
Design
The design of the study was a 2x2 two-way within subjects ANOVA. The dependent variable was measuring reaction time and the two independent variable (a) typicality of object colour and (b) adding blurriness to both typical and atypical images within participants.
Materials
The experiment used computer-run programme called E-prime script. Coloured pictures of everyday objects were obtained from Naor-Raz et al. (2003) original experiment and different stimuli form TarrLab (http://wiki.cnbc.edu/TarrLab). Total of 80 pictures were used using 20 pictures in each condition (typical-clear, atypical-clear, typical-blurred, atypical blurred). To modify pictures for the second independent variable (blurriness), Programme Irfanview 4 was used. In contrast to Naor-Raz et al. (2003) original experiment, participants did not spoke the colour. Instead, coloured stickers of the seven basic identifiable colours were places on the keyboard so that participants would press the right key. The seven basic colours were; blue, red, green, purple, white, orange, and yellow.
Procedure
A single image of an object was presented in the centre of the screen on each trail. Participants were instructed to press the key of the relevant colour that is shown on the screen as quickly as possible as our dependent variable was measuring reaction time, and also they were told to ignore the object it self. Pictures were remained on the screen until the participant responded by pressing the key. All participants took part in the experiment in St Johns University computer lab with a very quite environment. Although all the participants were doing this experiment as a part of their course, every individual was given a reference number on their consent sheet to protect their confidentiality and were told about their right of withdrawal at any point during the experiment, and to email the project supervisor in case they wanted to withdraw their data later.
After the experiment the data was analysed using 2x2 within subjects ANOVA using the programme SPSS
Results
As table 1 shows, participants in the typical-blurred condition had the fastest mean reaction time, and participants in the atypical-blurred condition had the lowest mean reaction time.
Kolmogorov-Smirnov test were used to examine whether the data were normally distributed and found that the data for atypical-blurred condition, D(20) =0.08, p=ns, atypical-clear condition, D(20) =0.16, p=ns, typical-blurred condition, D(20) =0.06, p=ns, and typical-clear condition, D(20) =0.17, were all normally distributed. Therefore differences in conditions were examined using a 2x2 dependent (within-subject) ANOVA.
The results found a non-significant main effect for the typical colour condition F(1,19) =4.258, p=ns. There was a non-significant main effect for the atypical colour condition F(1,19)= 0.629, p=ns. There was a non-significant interaction F(1,19)=1.935, p=ns.
Discussion
The above results show that there was no significant effect across the different conditions (typical-clear, atypical-clear, typical-blurred, and atypical-blurred) and by using the 2x2 ANOVA there was no significant interaction. Unlike the findings of Naor-Raz et al. (2003) where they found a significant difference between the different colour conditions which indicated in their experiment that colour must be associated at some level of representation with the object’s shape. The non-significant results of this experiment also differ from Tanaka and Presnell (1999) findings that colour does play a role in the recognition of high diagnostic colours. This suggests that there are some mixed evidence of the previous research by Naor-Raz et al. (2003) and Tanaka and Presnell (1999) and the findings of the current experiment. In addition, even after adding another variable (blurring), the results would remain constant to the results of the previous experiments conducted by Naor-Raz et al. (2003) and Presnell (1999). However, given that there is mixed evidence from this experiment and from the previous ones, it only suggests that colour perception differs according to individuals and apparent conditions.
Primarily, the question being addressed in this study is about whether colour plays a role in object recognition specifically whether colour assists in object recognition. Based from the performed and past experiments, the benefit of colour potentially arises in two different ways. One when the colour is present in the visual display and the other is when colour is part of the stored knowledge that can be recalled in the absence of a representing object. It appears form the findings that perceptual colour still remains inconclusive despite the amount of experiments and studies conducted on both normal and conditional subjects. There is also an emerging consensus that stored knowledge about the colour is subsequently important in identifying objects or colour. Together the results suggest that even though surface colour is seemingly part of a perceptual salient dimension, recognition cannot still be relied upon the presence of colour alone.
Despite the variation of results from the recent experiments and of this study, it can be assumed that the colour still plays a significant role in object recognition. Marr and Nishihara (1978) presented their argument in recognition supporting Naor-Raz et al. (2003) findings that colour recognition can be achieve using object representation. According to Marr and Nishihara (1978) visual shape recognition is based on the representation of shape. Similarly, Naor-Raz et al. (2003) incorporated the same findings in proving colour recognition. The only difference is the approach used by Marr and Nishihara (1978) because they focused in utilizing objects for recognition while Naor-Raz et al. (2003) used the object representation principles in colour recognition. It is one of the characteristics that set this research apart from the previous one. Such difference creates an impact to the findings of this study in proving the relatively important role of colour in recognition without the apparent need for a representing object.
In a more in-depth analysis of the results, images may be recognized faster particularly because the objects contains characteristics or information about a familiar shape that are stored in memory. Higher reliance in colour shows an existence of surface attributes. In relation to the responses from the samples, the subjects tend to recognize the images faster because they already have a familiar knowledge of the image and the colours attributed to the known objects provides a significant clue in recalling the objects and the colour of perceived from the images. Although the results are not consistently similar to the findings of the modelled experiment by Naor-Raz et al. (2003), it can still be assumed that the mixed results still shows the desired outcome and portion of similarity to the Naor-Raz et al. (2003) experiment, which aims to measure role of colour in object recognition.
References
Biederman, I., & Ju, G. (1988). Surface versus edge-based determinants of visual recognition. Cognitive Recognition, 20, 38-64.
Biederman, I. (1987). Recognition-by-components: A theory of human image understanding. Psychological review, 94(2), 115-147.
Curran, T., Tanaka, J. W., & Weiskopf, D. M. (2002). An electrophysiological comparison of visual categorization and recognition memory. Cognitive Affective & Behavioral Neuroscience, 2(1), 1-18. doi:10.3758/CABN.2.1.1
Logothetis, N. K., & Sheinberg, D. L. (1996). Visual object recognition. Rev Neurosci, 19, 577-621.
Marr, D., & Nishihara, H. K. (1978). Representation and recognition of the spatial organization of three-dimensional shape. Proceedings of the Royal Society of London. Series B, Biological Sciences, 200(1140), 269-294.
Marr, D. (1982). Marr, David (1982) Vision. San Francisco: W.H. Freeman. Retrieved from http://www-rci.rutgers.edu/~cfs/472_html/CogArch/DM.html
Naor-Raz, G., Tarr, M. J., & Kersten, D. (2003). Is color an intrinsic property of object representation?. Perception, 32, 667-680. doi:10.1068/p5050
Tanaka, J. W., & Presnell, L. M. (1999). Color diagnosticity in object recognition. Attention Perception & Psychophysics, 61, 1140-1153. doi:10.3758/BF03207619
Vuong, Q. C., & Tarr, M. J. (2006). Visual Object Recognition. Perception, 0, 1-41. doi:10.1068/p5491
Yantis, S. (Ed.). (2001). Visual perception: Essential readings. Philadelphia, PA: Psychology Press.
