week6_lecture2_scribe3 (Chris Crawford)

PSYCHOPHYSICS
Week 6, Lecture 2: Psychophysical methods
What is Psychophysics? • Quantitative approach towards understanding perception • Consists of a class of methods that can be used to study a perceptual system • Scientific study of the relationship between stimuli, sensations, and perceptions • Psycho o Mental & subjective • Physics o Physical measurements & objective • Psychophysics examples Psycho Physical Lengths Size Weight Heaviness Intensity Brightness Psychophysics Methods / Tools Perceptual quantities cannot be measured via a physical scale. Instead they are measured through inferred responses from people. A class of methods / tools quantify the relationship between physical entities and the perceptual response from people interacting with them. This could also be described as people’s experience while interacting with various physical quantities. Real Life Examples Much work in virtual reality (VR) involves the perception of size. For example, in a virtual environment the perceived size of a building, car, or doorway varies based on many factors. Distance is one factor that influences the perceived size of objects in a virtual environment. Although the measurements could be the same the environment is also a factor. Perceived size and distance in a VR environment differs from those same quantities in the real world. Heaviness is a physical quantity that is used with haptic. An example of a contributing factor with this example is texture. Haptic is used with touch in the same way that computer graphics is used with vision. It could be used to provide user’s feedback during activities such as gaming. This feedback is usually linked to an event that is communicated to the user visually. Brightness is a physical quantity used commonly in displays. It measures the intensity of light and color. Many displays allow user’s to change the brightness of the screen if desired. 2 Not above absolute threshold Above absolute threshold Fig. 1. Image of black boxes not above and above absolute threshold Fig. 2. Two black boxes with varying intensities Absolute Thresholds Absolute threshold is the intensity at which a physical stimulus becomes recognizable to an observer. One example of this can be observed with sound. Consider a sound begins at a volume where it cannot be heard. Now consider that the volume is being incremented by one unit recursively. The volume level at which the sound is first observed serves as the absolute threshold for sound. Another example of absolute threshold can be observed through brightness. The left side of figure 1 features a black box that is not visible. This is an example of an object that is not above the absolute threshold. The black box on the right of figure 1 is faintly visible in result to being right above the brightness absolute threshold. Absolute thresholds are measured through detection tasks. An example of this task is manipulating the intensities of the two black boxes in figure 2. To discover the absolute threshold responses must be gathered based on the question ‘What is the difference in intensities at which observers can notice that the two boxes are different’. Difference Thresholds Difference threshold is the amount of change necessary to cause observers to perceive a difference in a signal. This is also called a just noticeable difference (JND). An example of this could be experienced when someone is adjusting the volume to a radio or television. Often times even though the volume is being adjusted user may not perceive any difference. In result the user continues to adjust the volume until it reaches the desired level. The point at which the observer is able to notice change between two intervals represents the difference threshold. Discrimination task are used to retrieve difference thresholds. Scales Scales measure the change between two physical quantities. Scales address questions such as “Is this twice as bright?” or “Is this twice as loud”. Fig. 3. Formula for just noticeable difference Fig. 4. Weber’s law formula [1] Table. 1. Values computed using Weber’s law values ΔI 0.1 1 2 Kw 0.1 0.1 0.1 I 1 10 20 Table. 2. Example ascending and descending trials Trial # 1 2 3 4 … Series D A D A … Starting 135 111 138 115 … Intensity Final (A) 125 131 … Ma
Md
Final (D) 126 121 … Weber’s & Fechner’s Work Most of the preliminary work in this area was done in the 1800s. Ernst Heinrich Weber and Gustav Theodor Fechner were key in many of the early developments. Weber was interested in studying people perception of heaviness. During his experiments he had people come in and lift weights. The goal of these experiments was to investigate whether participants could tell the difference between two weights. Weber was interested in quantifying the experience of lifting something. After participants would lift two weights Weber would inquire whether they could tell the difference. These discrimination tasks could be used to compute the JND. For example if a participants noticed a difference between weight 1 (w1) and weight 2 (w1) then the JND could be computed using the formula shown in figure 3. Weber discovered that the JND was proportional to the weight itself. As Weber gave participants heavier weights he needed to make them further apart before they could tell the difference. This became known as Weber’s law. Weber law states that the just perceivable difference in intensity is proportional to the intensity itself. Figure 4 shows the formula for Weber’s law that upholds this principle. In the formula ΔI represents the change in intensity. Kw is a constant that is different for varying sensory modalities. ‘I’ holds the intensity. Table 1 shows computed values assuming Kw is equal to 0.1. Notice that as ‘I’ increases the change in intensity needed increases also. This means that the larger you make the base intensity the larger you need to make the change in order to perceive difference. Fechner was interested in measuring sensation. He stated that in order to measure subjective sensation a zero and incremental unit are needed. An example of a zero is the temperature at which water freezes. An incremental unit example is Celsius. The zero for a subjective measurement is the detection threshold. This is the threshold at which you can perceive a signal from no signal. The incremental unit is the JND. The JND changes as the intensity changes even though it still represents one unit. Fechner introduced methods for measuring these thresholds. An example of applying one method is through descending and ascending trials. Consider there are two squares with one being the standard and one being the test. Along with these squares is a dial that controls the intensity of the test square. Over multiple trials participants adjust the test square until it matches the standard square. This is done in a descending and ascending order. Table 2 shows an example of ascending and descending trials results. Computing the mean of the Final (A) row and Final (D) row results in Ma and Md. If Ma and Md are the same or similar there is no bias, otherwise if the two values are significantly different there is a bias. These variables can also be used to compute the grand mean. The formula for the grand mean is M = mean (ma, md). This grand mean also correspond to the point of subjective equality (PSE). The PSE measures the intensity needed for an observer to notice a difference. The PSE is used to compute the upper level (UL) and lower level (LL) threshold. The formula for upper level is UL = PSE + JND. The lower level formula is LL = PSE-­‐JND. These variables are used to compute the Interval of uncertainty. The interval of uncertainty is the interval at which two intensity are perceived as being equal. This is a very important principle in color science. An example of when this is useful when manufacturing paint. It would be a waste to produced two tones that fall within the same interval of uncertainty therefore this is avoided. Fig. 3. Example internal response probability of occurrence curves [2] Signal + Noise Noise Yes Hits Probability (Yes / S+N) False Alarm False Positive No Miss Probability (NO/ S+N) True Reject Fig. 4. Stimulus response matrix Signal Detection Theory Testing signal detection theory involves a two alternative forced choice test. The choices consist of whether someone observed a signal or not. An example of this is the task of a radar operator. Based on visual feedback from the radar the operator must make a decision of whether a threat is present on the radar or not. The task also involves filtering out noise. Figure 3 and figure 4 shows an example of how signal detection theory applies to this task. The region label hit represents when the operator correctly detected a threat. The miss region represents when the operator does not detect a threat that is present due to noise. A correct (true) reject occurs when the operator correctly responds no to the question of if a threat is present. A false alarm occurs when the operator detects a threat that’s no there. The criterion response represents when the operator says no (left of criterion) and yes (right of criterion). This could be affected by the result of the operator’s response. For example, if the yes response could result in high civilian casualties the operator criterion may shift due to avoid an accident. References 1. http://www.cns.nyu.edu/~msl/courses/0044/handouts/Weber.pdf 2. http://www-­‐psych.stanford.edu/~lera/psych115s/notes/signal/