Chapter 3 Cross-modal perception and applications to rendering
3.2 Selective attention
3.2.1 Visual attention
A broad distinction of attention in the visual domain can be made between the
Bottom-up ([Jam80]) (also known as exogenous orpassive) attention and the Top- down [DD95] (also known as endogenous oractive) voluntary shifting of attention. The former form of attention refers to the involuntary gazing to potentially conspic- uous stimuli in the environment, for example a car crash in front of us can definitely attract attention. The top-down attention refers to the voluntary focus of attention towards important visual stimulations not because they are salient but because they are considered as a“goal”. For example, the top down mechanism allows to read a magazine while sitting in a crowdy place. Top-down and bottom up are comple- mentary perceptual mechanisms and they do not work in an exclusive fashion. If attention was working solely in a bottom-up way then humans would never develop a behaviour directed by aims as salient stimuli would constantly distract them. On the other hand, even when focusing on a target, our attention still requires to be attracted by salient features of the surrounding environment.
Many researchers have reviewed the di↵erent characteristics of the bottom- up and top-down forms of perceptual attention [IK00]. On the one hand, bottom-up attention cannot be restricted, it is not a↵ected by person’s expectancies and previ- ous memories experiences and does not result to conscious awareness. The top-down attention model, on the other hand, is resource limited, it can be easily manipu- lated and it is strongly a↵ected by a person’s previous memories and expectancies. Top-down attention results in conscious awareness of visual cues.
Yarbus, in 1967 [Yar67], conducted a series of experimental studies to in- vestigate the e↵ect of top-down perceptual attention compared to the bottom-up
approach. Specifically, he used an eye tracking device to record participants’ eye trajectories when viewing an image (a copy of Repin’s painting titled “The unex- pected visitor”) under di↵erent experimental conditions. Subjects were asked to view a picture while they were assigned a specific task related to the content of the picture (e.g. to determine the ages of the characters that are depicted on the image or remember their clothes). Apart from the experimental trials that included a task, there were trials where participants were simply asked to freely examine the painting. The eye tracking results demonstrated that participants fixate on relevant objects when assigned a task compared to the case of freely viewing the image. This experiment clearly showed the distinction between top-down and bottom-up models of attention.
Visual attention can also be discriminated intoOvert andCovert shifts of at- tention [Pos80]. Overt shifts of attention describe the redirection of the eyes towards areas of interest or salient objects. The human eye usually performs overt shifts us- ing saccadic eye movements and fixation time intervals. On the other hand, covert attention does not involve eye movements and it enables faster response to attended visual events. Overt attention di↵ers from covert attention in the same way that gazing at an object is di↵erent than observing something with our peripheral vision (the corner of the eye). Experimental studies have measured reaction times (RT) in both covert and overt shifts of attention in either uni-modal scenarios or cases where multiple senses are delivered at the same time (multi-modal scenarios) [WW08].
Another psychological phenomenon that frequently occurs in the visual do- main is a person failing to notice an unexpected stimulus that is in plain eye- sight [RLGM92]. This phenomenon is known as Inattentional Blindness and it is not caused by physical inadequacies of the human eye but rather it characterises the failure of perceptual attention when watching multiple visual stimuli. This phe- nomenon was discovered and named by Mack and Rock in 1998 who claimed that the stage of consciously perceiving a stimulus requires first to attend it [MR98]. Specifically, the authors argue that the perception of a shape when watching a vi- sual stimulus requires visual attention while the perception of colour, location and motion features do not necessarily require attention to be consciously perceived. In another study, by Simmons et al. [SC99], the authors demonstrate a practical example of Inattentional Blindness through a task where participants watch two basketball teams (black and white) in a video to change ball passes. The task is to count how many passes the white team made. During this task, participants miss to attend to a gorilla or a woman with an umbrella passing through the scene. In the first demonstration of the experiment only 8% of the participants were able to
report the presence of the unexpected character in the video.
An equally significant phenomenon that is related to human perceptual at- tention is known asChanged Blindness. In this case, the HVS fails to notice visual di↵erences that are included in a visual scene or a digital image. Typically, for changed blindness to occur, the change in the scene needs to coincide with a visual distraction, for instance, a focus of attention to a di↵erent spatial position [ROC97]. The HVS does not capture the newly included information and the di↵erences re- main undetected by the viewer.