Year 2 Quantitative Report: The Effect of Different Categories of Change on Change Blindness
Note to the reader: Please remember that the essay is published work and partial reproduction (5 or more consecutive words) might be considered plagiarism. Universities have systems in place to ensure that the work of their students is original.
Grade awarded A+
The report does not include the table of contents, consent form, the debrief form and the SPSS files.
The Effect of Different Categories of Change on Change Blindness
Abstract (207 words)
Change blindness is the failure to notice when a change is made to a scene. Using the flicker paradigm, the present report tried to determine that the average response time to detect a change varies significantly depending on the type of change made. An opportunity sample of 250 individuals (180 females and 70 males) took part in a within participants design experiment. There were three experimental conditions characterised by the type of change made to a scene: an object was replaced by an item from the same category, from a congruent category and from an incongruent category. In the control condition no change was made. The results of the experiment did not find a significant difference in the average response time to detect a change in all three categories. Also, differently from what was predicted, the average response to detect a change was fastest in the congruent-change category and slowest in the incongruent change category. The use of more control measures for the object changed in each scene (such as colour, shape and location), and for the types of the scenes (complex, less complex) as well as for the demographic characteristics of the participants (equal number of males and females) could lead to more significant results.
Introduction (500 words)
The new research in cognitive psychology challenged the pre-existing belief that humans have an accurate and objective representation of the physical qualities of the environment. The discrepancy between the way the surroundings are perceived through the anatomical structures of the eye and the mental representation of the environment was studied using the concept of ‘change blindness’, explained as the failure to identify that a change is made to a scene (Briggs & Davies, 2015). Also, even when people are able to detect the change, central elements are easier to notice than marginal ones. Using the ‘flicker paradigm’, Rensink et al. (1997), cited in Briggs & Davies (2015) made an experiment where a blank was introduced between an alternation of a scene and a modified scene. On average, participants needed 5 seconds to find a central change in the scene and 11 seconds to observe a marginal change. Similar conclusions were reached by O’ Regan et al. (1999), cited in Briggs & Davies (2015).
Moreover, the phenomenon of ‘change blindness’ was also supported by experiments with high ecological validity. In an experiment conducted by Simons and Levin (1998) cited in Briggs & Davies (2015) 47% of people failed to notice that someone asking for directions is replaced by one of two people who are transporting a door. When students were asked for directions by a person dressed as a construction worker, those noticing a change of persons dropped to 33% and the explanation is that the students payed less attention to the workers because they belonged to a different social group. This shows that the quality of attention is important to detect when a change is made in an environment.
Other factors might predict change blindness. In one study, Miyamoto et al. (2006) cited in the University (2018, b) used an experiment to demonstrate that the attention can be even influenced by the cultural environment. Participants, either Japanese or Americans, who were first exposed to Japanese scenes (considered to be more complex) as opposed to American environments, were better at detecting changes made to a scene. Therefore, individual characteristics can influence the capacity to detect a change in a scene.
The current study investigates whether the average response time needed to detect a change in a scene can be determined by the type of change made. The changes are from the same category (within-category change), from a similar category (congruent change) and from an incongruent change. The change blindness research showed that the ability to notice a change is influenced by how salient objects are, as a central change is easier to detect that a marginal one. The reasoning behind the present experiment is that objects can be more or less salient, depending on which category they belong to. As a result, a significant difference between the average response time to detect a change between the three categories, with the fastest response time for the incongruent change category, followed by congruent change category and lastly, within category change is expected.
Method (770 words)
Design
The design is a within participants experiment that researched how different types of changes made to a scene (independent variable) determined different reaction times to detect that a change was produced (dependent variable). The independent variable has three conditions: 1) congruent change, 2) incongruent change, 3) within category change. The dependent variable is measured as the mean reaction time to detect a change for each of the 3 categories. To control that the participants correctly identify when a change takes place within a specific category, a quarter of the 16 everyday scenes don’t present any change. In this way the data of the people who frequently detect a change in the no-change trials can be removed before the analysis.
Participants
250 People took part in the experiment, with ages between 18 and 78. Over two thirds (180 out of 250) of the participants were females and just under a third of the participants (70 out of 250) were males. The average age was 37.1 years (with a standard deviation of 12.483) and half of the those who took part in the experiment were younger than 34 and a half years. Only 7 participants were over 60 years old. Every participant gave informed consent. The sampling was random, but the participants confirmed they had normal eyesight and that they were not psychology students of the DE200 course.
Materials
The experiment was conducted in a laboratory that was equipped with a desk, a chair and a computer with internet access. The experiment was accessed using the module (University, 2017 a). The instructions before the experiment started, the consent form and the debrief were included in the data file with the running experiment. Also, for convenience, the assistant had separate forms for experiment instructions, consent and debrief. Every participant took part in the experiment individually. The stimuli were randomly presented on the computer screen and they consisted of 16 experimental trials depicting everyday scenes. Each trial in the experimental conditions consists of two very similar images presented one after the other with a blank screen in the middle. The blank screen and the images appear as flickering on the computer screen for 0.25 seconds each. When the participants identify that there is a change between the two images, they click the image and the time taken to identify the change is recorded.
There are three types of changes made to the scene: ‘within-category change’- when an object is changed with one from the same category (when a phone is replaced by a stapler), ‘congruent change’- when the object is not from the same category, but fits in the context (a phone is replaced by a banana), and ‘incongruent change’ when the object is replaced with one that does not match in the context (a plate is changed with a watering can). The three types of changes appear four times each, in random order. The control condition consists of 4 trials and is similar with the experimental condition only that the images that flicker after each other are identical. The purpose of the control condition is to allow the researchers to remove the answers of the participants who consistently identify a change.
Procedure
Every participant was tested individually and before the experiment began, he/she was told that he/she could withdraw at any time. Also, every participant was asked to confirm that he/she had normal eyesight and in the case of having photosensitive epilepsy, the participant was warned to not take part in the experiment. After that, the purpose of the experiment was explained. Before the experiment started, the age and gender, as well as the consent to take part in the experiment were recorded in the date file containing the running experiment and also on the forms provided by the assistant. Also, the participant had the opportunity to see three examples with the requirements of the experiment. After that, the participant took part in the 12 trials in the experimental conditions and in the 4 trials in the control condition. Each trial contained everyday scenes that were slightly different in the experimental condition and identical in the control condition. The participants saw the trials in random order, and they had up to 20 seconds to click on the image if the trial depicted a change in the scene. After watching the 16 trials, the participants had to submit their responses. In the debrief form, the participant was thanked for the time he/she took to complete the experiment and was reassured by being told that around 30% of people fail to notice changes in similar trials.
Results / Analysis(312 words)
On average, the 250 participants’ response time was fastest for the ‘congruent changes category’ (mean= 4.29, SD=1,39), followed by ‘within category changes’ (mean=5.13, SD=2.00) and lastly, the slowest response time was for the ‘incongruent changes category (mean= 5.34, SD= 1.98)
Average response time (RT)
Mean
Std. Deviation
N
Average RT for Congruent Changes
4.2951
1.38711
250
Average RT for Incongruent Changes
5.3411
1.98698
250
Average RT for Within Category Changes
5.1306
2.00827
250
The one-way repeated measures ANOVA determined that the effect of the type of change made to a scene (IV) on the mean reaction time to notice that a change was made (DV) was statistically significant (F (1.78, 444) =33, 84, p<0.001, np2 = 0.12) Partial eta squared (np2) measures the effect size. np2 = 0.12 which means that the independent variable (the type of change made to a scene) has a medium effect on the dependent variable (the mean reaction time to notice the type of change). Mauchly’s test of sphericity is statistically significant meaning that the condition of sphericity has not been met because p<0.005 χ2 (2) =0.88, p=0.00. To correct the sphericity the Greenhouse-Geisser estimate was used. The mean difference is statistically significant (Sig.b < 0.001) between congruent change made category (1) and within category change made category (3) and also between congruent change made category (1) and incongruent change made category (3). However, the mean difference is not statistically significant (Sig.b = 0.51) between incongruent change category (2) and within category change (3). The findings do not support the hypothesis that there is a statistical difference between the mean response times to identify a change in the three categories. Also, contrary to the hypothesis, the results show that the average response time for the congruent changes is fastest for the within-category changes and it is the slowest in the case of incongruent changes category.
Discussion (585 words)
The purpose of this study was to research how change blindness affects people depending on different types of changes made to a scene. Data analysis did not support the hypothesis and contrary to what was expected the response time to identify that a change was made to a scene was slowest in the case of incongruent change and fastest for the congruent-change category. The outcome is surprising considering that the items that replaced other objects in the incongruent change category were more salient, for example in one of the paired scenes, a white pillow is replaced by a white bin.
The result could partially be explained by the lack of controls for the shape, colour and place of the objects chosen for the three experimental conditions.For example, in the within category change, the objects are changed by items from the same category: a mug with striped design is changed with another mug with three big dots. Of course, both mugs belong to the same category, but arguably, the three big dots draw attention and it makes it easier for the participant to identify a change. Moreover, in the present study, the objects changed are randomly positioned in the centre or more marginally in the scene. Previous research, (Rensink et al. (1997), cited in Briggs & Davies (2015) and O’ Regan et al. (1999), cited in Briggs & Davies (2015) demonstrated that central changes are easier to detect than marginal ones. Thence, the salience of an object from the within-category change (most difficult to detect), is increased if the object is placed in the centre of the scene.
Another problem might have arisen from the degree of complexity of the scenes, as some of the images included in the experiment contained more details than others. In the study of change blindness, Miyamoto et al. (2006) cited in the Open University (2018, b) used images with different degree of complexity (Japanese vs American environments) to assess that the cultural environment has an effect on change blindness. Although all the images in the experiment are everyday scenes depicting places in an office, bedroom or kitchen, some of the images included in the present study contain more elements than others and this can have an influence on the speed of identifying a change in the three categories.
The current study could be improved by using controls for the colour, shape and location of the objects changed in each scene. Also, it could be useful to control variables such as age and gender. Less than a third of the participants included in the experiments were males and the evidence (Cassimjee & Maree, 2004) suggests that males and females are differently affected by change blindness (including changes in colour and location of an item).
Even if the current study failed to achieve significant results, it indicates that the quality of an object to belong to a certain category is not enough to obtain a significant result in the average response time to detect a change in different categories of change. A replication of the experiment using more control measures could lead to different results. Still, there is no certainty in the way cognitive processes other than perception, attention and memory can facilitate object detection in change blindness. For example, one study (Porubanova-Norquist & Sikl, 2013) suggests that participants have different expectations about the location of the change depending on the elements that attract attention in a scene and in order to optimise the detection, they use a heuristic approach.
Total word count (without the abstract): 2175 words