Thresholds, Weber’s law, Fechner’s three methods Research Methods Fall 2010 Tamás Bőhm Introspection • Method applied by early psychologists (Wundt) • Self-observation of conscious thoughts and sensations – Do I perceive this light brighter than that? Why? • Inherently subjective rejected as a scientific method • But it provides useful intuitions when designing experiments (helps forming hypotheses) – I realized that it is hard for me to hear very high and very low tones. Thus the frequency of a tone may be a factor in its perceived loudness/audibility. Let’s test this in an experiment! Formal methods • Quantifying the sensations evoked by physical stimuli • Gustav Fechner (1860): established 3 formal psychophysical methods • Fechner’s methods have been (and are being) widely used • More reliable and accurate than introspection • Results from different experiments are comparable Sensation magnitude Thresholds If a linear relationship is assumed, two values determine the function: Linear psychophysical equation X-intercept slope Stimulus intensity – X-intercept: minimum stimulus value that evoked any sensation; absolute threshold – Slope: the rate at which sensation grows as we increase intensity; difference threshold (inversely proportional to slope) Thresholds General definitions (not assuming linearity): – Absolute threshold: intensity that the observer can just barely detect • Intensities below absolute threshold: undetectable • Intensities above absolute threshold: detectable – Difference threshold (aka. just noticeable difference /JND/ and difference limen): minimum intensity difference that is noticeable to the observer • A change in intensity that is smaller than the difference threshold: undetectable • A change in intensity that is larger than the difference threshold: detectable Difference thresholds Sensation magnitude • Linear psychophysical equation • constant slope Stimulus intensity • • • • Linear function difference threshold (slope) is constant An observer able to detect the difference between intensities 100 and 110 should also be able to detect the difference between 1000 and 1010. This is not the case: the observer is able to detect the difference only between 1000 and 1100 500 & 550 Hz tones 5000 & 5050 Hz tones 5000 & 5500 Hz tones Difference threshold is not constant! Difference thresholds Sensation magnitude • Nonlinear psychophysical equation slope changes with intensity • • • Stimulus intensity Difference threshold is not constant (changes with intensity) function is nonlinear Weber’s law: difference threshold is a constant proportion of the initial stimulus value ΔI / I = c Previous examples: c=10% Weber’s law holds only approximately! Absolute thresholds • Even in the absence of stimulation, there is some random firing on sensory nerves • This inner noise can even vary from moment to moment • Observers cannot distinguish inner noise from the effect of a weak stimulus • Even when there is no light (perfect darkness), observers may experience a dim light (dark light, intrinsic light) • Observers in an anechoic chamber often report hearing a whistling sound Measuring truly „absolute” thresholds is problematic: observers may confuse inner noise with the real thing Psychophysical methods • Threshold measurements: detection of small intensities (absolute thr.) and discrimination of small intensity differences (difference thr.) Is it intense enough to see? How small a difference can you see? – Fechner’s 3 methods • Method of constant stimuli • Method of limits • Method of adjustment – Modifications of Fechner’s methods • Staircase method • Modifications of the method of constant stimuli (adaptive, no standard) – Forced choice, objective methods – Sensory decision theory (SDT) – Psychophysical functions from psychometric data • Direct scaling: growth of sensation with intensity How bright do you see a light? – Magnitude estimation and the power law • Multidimensional scaling: degree to which stimuli are comparable along some dimensions Along which dimensions do you judge the similarity of two stimuli? Fechner’s three methods • • • Absolute threshold Difference threshold Method of constant stimuli Method of limits Method of adjustment not used • Presenting one stimulus at a time The stimulus is very weak Possible responses: “Yes, I see it.” / “No, I don’t see it.” • Presenting two stimuli at a time: – Standard: fixed, easily detectable – Comparison: either more or less intense than the standard Possible responses: “Comparison is stronger.” / “Comparison is weaker.” Method of constant stimuli for measuring absolute thresholds Light intensity Weak Strong 1. Select a range of light intensities from certainly invisible to certainly visible 2. Pick a few (4-7) points uniformly in this intensity range; this will be the constant stimulus set Method of constant stimuli for measuring absolute thresholds … 3. Test each stimulus many times (20-25) in random order Method of constant stimuli for measuring absolute thresholds Visible? YES NO 4. Present the stimuli one at a time and ask the observer if it was visible or not Click to start Could you see the spot of light? Could you see the spot of light? Could you see the spot of light? Could you see the spot of light? Could you see the spot of light? Could you see the spot of light? Could you see the spot of light? Could you see the spot of light? Could you see the spot of light? Could you see the spot of light? Could you see the spot of light? Could you see the spot of light? Could you see the spot of light? Could you see the spot of light? Method of constant stimuli for measuring absolute thresholds 0% + - 5% + + + 20% - - 50% - + 80% + + 95% 100% - - - - 5. Calculate the proportion of “yes” and “no” responses at each light level + Percentage “seen” Method of constant stimuli for measuring absolute thresholds 100% 75% 50% 25% 0% Stimulus intensity 6. Plot the percentages against stimulus intensity psychometric function Psychometric function for absolute thresholds Ideal FIG (Sekuler) • Fixed absolute threshold • Step function Actual sigmoid function FIG (Sekuler) • Absolute threshold varies somewhat from trial to trial (due to constant fluctuations in sensitivity) • Conventionally, the intensity corresponding to 50% is considered to be the threshold Method of constant stimuli for measuring difference thresholds Standard stimulus: Comparison stimuli: Light intensity 1. Standard stimulus has a fixed intensity 2. The intensities of comparison stimuli bracket the standard Method of constant stimuli for measuring difference thresholds 3. All pairs of standard and comparison stimuli are tested many times Method of constant stimuli for measuring difference thresholds STRONGER WEAKER 4. For each pair, the observer judges if the comparison stimulus was stronger or weaker than the standard http://www.yorku.ca/psycho Percentage “stronger” Method of constant stimuli for measuring difference thresholds 100% 75% 50% 25% 0% Light intensity of comparison stimuli 5. For each comparison level, the percentage of “stronger” responses is calculated and results are plotted as a psychometric function Percentage “stronger” Psychometric function for difference thresholds 100% 75% 50% 25% 0% PSE Light intensity of comparison stimuli • When the observer cannot see a difference, he/she chooses randomly between “stronger” and “weaker”; this corresponds to 50% on the psychometric function point of subjective equivalence (PSE) Percentage “stronger” Psychometric function for difference thresholds 100% 75% 50% 25% 0% DW DS Light intensity of comparison stimuli • • • By convention, the intensity at 75% is considered to be just noticeably stronger than the standard DS A comparison intensity at 25% is just noticeably weaker than the standard DW Difference threshold = the average of DS and DW Method of limits for measuring absolute thresholds 1. + Light intensity + + 2. + + + - threshold estimate - 3. 4. Trials On each trial, the observer reports if she/he could see the light or not. Start with presenting a light intensity well above the expected threshold (the observer can certainly see it) Decrease the intensity until the observer cannot see it Threshold estimate: the intensity at which the response changes • Descending series: start from above the expected threshold and decrease intensity • Ascending series: start from below the expected threshold and increase intensity Click to start Method of limits for measuring absolute thresholds + + Light intensity • + + + + + + + - - + • + - - - - - - - Trials • Ascending and descending series may yield different results use both Even in the same direction, there is variability in the threshold (inner noise, etc) average many measurements Measured threshold corresponds to 50% point in a psychometric function (method of constant stimuli) pure-tone audiometry Light intensity of comparison stimulus Method of limits for measuring difference thresholds + + + + + + + threshold + estimate - + + + + - - - Trials + comparison brighter - comparison weaker • Intensity of the comparison stimulus is decreased (descending) or increased (ascending) until the response changes • Threshold estimate: intensity difference between the standard and comparison stimuli where the response changes • Average results from multiple series in both directions Method of adjustment for measuring absolute thresholds • Observer can vary the stimulus intensity • Instructed to adjust it so that it is just visible or just invisible • Initial intensity is set to be far from the expected threshold value adjustment device stimulus with adjusted intensity observer Method of adjustment for measuring absolute thresholds threshold estimate descending ascending • Adjustment: by a real or a software device (e.g. knob, slider) • Threshold estimate: final intensity value • Descending: initial intensity is well above expected threshold; adjusted to just visible • Ascending: initially well below threshold; adjusted to just invisible • Ascending and descending task repeated several times and results averaged • Similar to method of limits but observers find it easier