Reading...
Play button
Play button
Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
39 Cards in this Set
- Front
- Back
|
Frequency of Seeing (FOS) curve: psychometric function
|
Observer required to detect a light flashed on a background. Task repeated for a range of stimulus intensities , from DIM to intense; % of stimulus detected is plotted as a function of stimulus intensity
|
|
|
ideal observer
|
unambiguous threshold; below threshold, stimulus is never seen, and above, stimulus always seen
|
|
|
real observer
|
as intensity of stimulus increased, the PROBABILITY of seeing stimulus increases. NO clear defined intensity
|
|
|
Threshold
|
defined as intensity that results in detection of the stimulus on ONE-HALF of the presentations
|
|
|
Stimulus results in..
|
.NEURAL activity; if sufficiently strong, the stimulus is seen. RANDOM neural noise is inherent.
|
|
|
Diseased visual system causes FOS curve to...
|
be less steep, and more flat. b /c more difficult to accurately ascertain a threshold
|
|
|
Methods of measuring threshold: ASCENDING LIMITS
|
stimulus initially below threshold, NOT visible. stimulus intensity increased until observer reports visibility. results AVERAGED over trials. PROS: dark adaptometry
CON: predicting the stimulus over trials |
|
|
Methods of measuring threshold:
DESCENDING LIMITS |
start with clear VISIBLE STIMULUS, visibility decreased systematically. USED FOR VISUAL ACUITY. e.g. SNELLEN CHART. PRO'S: familiarize patient, clinician knows pt understands what to do. CONS: observer anticipation
|
|
|
Methods of measuring threshold:
STAIRCASE METHOD |
combination of ASC and DESC LIMITS. Repeat A and D, and then after one of the reversals, threshold is taken to be the stimulus intensity. Go from Not seeing to Seeing.
|
|
|
Methods of measuring threshold:
CONSTANT STIMULI |
STIMULUS visibility VARIED randomly from presentation to presentation. "YES-NO" procedure . BLANK trials also included (number of times observer reports seeing blank stimulus = false positive). CREATE chart that contains, HITS, MISSES, FALSE POSITIVES, CORRECT REJECTS. CONSTANT b/c maintains EXPECTATIONS at a constant level. CONS: time consuming, errors b/c subject is biased (too strict or lax threshold criterion). Make up for this by including SUPRAthreshold stimuli; when subj says does NOT see= false NEGATIVE. CORRECT for error on FOS curve
|
|
|
Methods of measuring threshold:
ADJUSTMENT |
subject adjusts intensity until it is barely visible allowing for quick threshold determination. CONS: ANTICIPATION and VARIATIONS in observer's threshold criterion
|
|
|
Methods of measuring threshold: Forced CHOICE method
|
Pt's may have low or high sensitivity to stimulus, and can vary trial to trial. FCM minimizes this. 1) FORCE them to choose between several choices, one of which contains the stimulus. e.g. One blank the other has the stimulus (threshold criterion reduced). 2AFC experiment. LOWEST % correct = 50%. Threshold midway between chance performance and perfect performance.
2) 4AFC experiment. chance performance= 25%. Threshold = 62.5%.4AFC more accurate/ threshold ascertained with more CONFIDENCE. STEEPER GRAPH |
|
|
Signal Detection Theory
|
Model to predict effects of decision criteria, attention, motivation, and internal neural noise. ASSUMES that always randomly fluctuating level of background neural activity. STIMULUS produces neural signal that is SUPERIMPOSED on neural noise. OBSERVER must distinguish signal and noise combination from the background noise alone.
|
|
|
NEURAL noise distribution
|
RANDOMLY fluctuating, BELL CURVE
|
|
|
The larger the stimulus...
|
Distributions N and N +S becaome further apart and DETECTABILITY of stimulus increases. NO uncertainty when there is no overlap.
|
|
|
Effects of observer criterion measurements
|
HIT: if stimulus results in neural activity exceeds threshold criterion. MISS: stimulus does not exceed threshold. FALSE positive: no stimulus but neural activity exceeds criterion. CORRECT REJECT: neural noise BELOW criterion during a blank trial (pt should NOT see in a blank trial)
|
|
|
Lax criterion in measurements:
|
FEW misses, MANY HITS, many false positives
|
|
|
STRICT criterion:
|
(d' is the same as lax criterion, only criterion line changes) MORE misses, fewer hits, few false positives, and more correct rejects
|
|
|
ROC curve (receiver operating characteristic)
|
shows probability of a hit as a function of the probability of a FALSE POSITIVE, thus, we are able to predict the effect of the observer criteria for a given detectability
|
|
|
ROC curve: when d'=0,
|
STRAIGHT LINE, stimulus is so weak that it produces virtually no signal; no matter what the criteria, proportion of HITS matches PROPORTION of FALSE POSITIVES (can't distinguish)
|
|
|
ROC: when d' = infinity
|
VERY CURVED: observers' criterion has no effect on proportion of hits and false positives. OBSERVER always sees the stimulus, and there are NEVER false positives
|
|
|
How to control observer's criterion:
|
STRICT: rewards for hits, and penalties for false positives
LAX: penalty for false pos minimized relative to the reward for a hit |
|
|
ROC curves: sensitivity vs. specificity
|
HITS vs. correct rejects. TRADE OFF between hits and correct rejects as we change threshold criteria. CUT OFF criterion needs to result in HIGH sensitivity and HIGH specificity. OPTIMIZE area of ROC curve= Better test.
|
|
|
discriminate between combo of stimulus and background and the background alone using what experiment?
|
increment threshold experiment
|
|
|
THRESHOLD INCREMENT
|
JUST NOTICEABLE DIFFERENCE (JND) or DIFFERENCE LIMEN (DL)
|
|
|
JND
|
changes as the background changes. as background INCREASES< the JND also increases such that the ratio of JND/background intensity remains CONSTANT
|
|
|
Weber's LAW?
|
JND/background intensity remains CONSTANT
Increment threshold JND = K Background illumination KI = JND/background illumination |
|
|
As background becomes more intense...
|
increment threshold increases, absolute sensitive decreases, while relative sensitivity remains constant
|
|
|
SCOTOPIC vision
|
vision of eye under low light conditions: contrast threshold= 0.14
|
|
|
PHOTOPIC vision
|
vision of eye under well-lit conditions: contrast threshold is 0.015
|
|
|
Sensitivity regulation
|
constant contrast threshold regardless of background brightness
|
|
|
lightness constancy:
|
visual system has evolved to detect contrast, rather than absolute luminance, so appearance of optotype is the same even with changing dim and bright lighting conditions. (b/c amt of light reflected on surface and background is the same)
|
|
|
Simultaneous contrast:
|
Darker background makes central square appear bigger. ; BRIGHTNESS DEPENDS ON BACKGROUND
|
|
|
Magnitude of Sensation
|
if an intensity of a light bulb is doubled, will it appear twice as bright? Relationship between magnitude of sensation and stimulus intensity
|
|
|
Fechner:
|
indirect scaling: assumed that ALL JNDs produce equal differences in the mag of sensation. INCORRECT
|
|
|
STEVENS
|
asked observers to directly assess the intensity of suprathreshold stimuli: DIRECT SCALING; observer asked to assign a number to the perceived brightness of each stimuli. VERY dim light = 1 very bright light = 10. Sensation plotte4d vs. light ntensity.
|
|
|
Stevens' power law
|
S = I ^c
S= mag of sensation I= stimulus intensity c= constant Growth in mag of sensation follows a power relationship rather than a log relationship |
|
|
What does Stevens' law show?
|
COMPRESSION (Saturation) of sensation as stimulus intensity increases. ( limiting factor). e.g. light bulb. 0, 50, 100, 150 W. physical diff between each setting is the same, yet turning the light on from 0 to 50 W is much more noticeable than adjusting the setting from 100 to 150 W.
|
|
|
FOS curve:
|
Letter size vs. % correct; frequency of seeing, steeper curve= more accurate. Very small within subject variance= very narrow gaussian curve= steep FOS curve
|
