1. Psychophysical Assessment
of Visual Function
As an O.D. you will measure (assess) vision.
How well does the person see?
Results depend on how you make the measurement

2. You: Any problems with your vision?
Pt: Don’t seem to see so well, Doc.
What do you do (how do you learn how well the patient sees)?

3. You measure the patient’s vision.
This course is about the science that stands behind why you measure vision certain ways in the clinic.

4. There are many different eye charts
Which chart to use?
How many letters per line?
How far apart are the letters and lines?
How much smaller are the letters on the next line?
Which letters to use?
How far down the chart must the patient try to read?
How score the result?
The acuity you get will differ depending these factors

5. It is a matter of judgment that determines how the visual system is tested and what constitutes normal variation in sensory processes.
The clinician must understand the scientific basis on which these judgments are made and how they can be made in the future as new tests of visual function are developed.
That’s what this course is about and why it is clinically relevant

6. Dr. Tom Norton
606 Worrell Building
Graduate Student Teaching Assistant:
Jason Wilson

7. Class – Mon- 9-10:50 Tues, 9-10:50 Wed, 11:00–11:50 Lab on 4 Thursdays
Check the schedule for your day and time
(Schedule will be distributed tomorrow)
This week: Group C 1-3; gp A 3-5

8. Exams #1 Wed. Jan. 19 (100 pts) Ch 1-3
#2 Tues. Feb. 8 (100 pts) Ch 4-5
Final during Final Exam Period (130 pts)
(110 pts new, 20 pts cumulative)
Labs (4 x 10 pts)
Possible pop quizzes (up to 30 pts total)
Total possible points, 370 (up to 400)
Letter grade end of course

9. Labs
Attend at the assigned day and time (unless you make other arrangements with Dr. Norton in advance)
Lab Reports due at Monday class after your lab
Accurately recording and graphing your data is an important part of your lab grade

10. Student-submitted exam questions
A way to control your own future!
Procedure:
Due several days before exam ( or Word files preferred)
Norton reviews, corrects, photocopies
Distributed to class (can use as a study guide)
Some of the questions will be used on the exam

11. Three main purposes of course
Learn how vision is measured (scientific basis)
Basic facts about monocular visual function
What is normal?
Neural basis of visual function
Why does the visual system respond as it does?

12. Textbook The Psychophysical Measurement of Visual Function
Norton*, Corliss, Bailey
Richmond Products, Inc
(*TTN’s author royalties [$ so far] donated to the UABSO)

13. We will cover 9 Chapters Principles of Psychophysical Measurement
Absolute Threshold of Vision
Intensity Discrimination
Adaptation to Light and Dark
Spatial Acuity
Spatial Vision
Temporal Factors in Vision
Skip Chapter 8 (color)
8) Postnatal Human Vision Development
The Aging Visual System

14. Overview At the beginning of each chapter.
Contains a summary of the content of the chapter.

15. Declarative section headings summarize the section they precede
“In the Method of Constant Stimuli the examiner randomly presents a set of stimuli with fixed, predetermined values”
“Correct for guessing by incorporating catch trials”

16. Study Guide
Questions at the end of each chapter intended to help you clarify your knowledge – (not as useful as I had hoped)
Lecture overlaps with the book a lot
… but questions also come from the book on topics I don’t cover in class!

17. Glossary – intended to help you know what terms mean for exam
Definitions given in the text – definite full credit if you know them verbatim
Equations – must tell what the variables mean

18. Equations – must tell what the variables mean
“What is the Stevens Power Function?”
where Y (psi) is the sensory magnitude, k (kappa) is an arbitrary constant determining the scale unit, F (phi) is the stimulus magnitude, and a (alpha) is an exponent that is characteristic of the stimulus used.

19. Graphs – The hardest part of this class
(because they tend to all look alike)
… but important because they show the relationship between stimuli and responses

20. Graphs – how to dissect and learn them
What is on the X-axis? (& approx. scale)
Usual arrangement:
Physical Stimulus on X-axis (Independent Variable)

21. Graphs – how to dissect and learn them
What is on the X-axis? (& approx. scale)
What is on the Y-axis? (& approx. scale)
Usual arrangement:
Response on Y-axis
What you are measuring
(Dependent Variable)
Physical Stimulus on X-axis (Independent Variable)

22. Graphs – how to dissect and learn them
What is on the X-axis? (& approx. scale)
What is on the Y-axis? (& approx. scale)
How plot a data point?
Usual arrangement:
Response on Y-axis
What you are measuring
(Dependent Variable)
Physical Stimulus on X-axis
(Independent Variable)

23. Graphs
What is different in each graph in a “family” of curves?

24. Lots of details to learn.
Philosophy: better to have learned and forgotten than to not have learned in the first place.
example

25. “Joke break”
Break the monotony
… but remember that the course has a serious purpose, and the exams can be difficult.

26. Student Response System
Test to see if it works
Will use for feedback
Will not look at who responds
Set to room code (23)

27. Principles of Psychophysical Measurement
Chapter 1
Principles of Psychophysical Measurement
Objectives:
Psychophysical Methods
Threshold
Constant Stimuli
Limits
Adjustment
Signal detection theory
Sensory Magnitude

28. We study visual psychophysics, but there also is auditory psychophysics, somatosensory psychophysics, etc.

29. Why are there so many graphs in this course?
Because graphs show relationships
Usual arrangement:
Response on Y-axis
What you are measuring
(Dependent Variable)
Physical Stimulus on X-axis
(Independent Variable)

30. of psychophysical measures
Two basic types
of psychophysical measures
Threshold measures (Do you see it”)
2) Sensory Magnitude measures
(“What does it look like”)

31. Do you see the light?
Physical stimulus – light intensity
Perceptual response – Seeing the light

32. How far down an eye chart can you read?
Physical stimulus – Letter size
Perceptual response – Identifying letters

33. letter size is the stimulus
identifying letters is response
We use psychophysical tools to find the threshold – the letter size you can see 50% of the time

34. Which is better, 1 or 2?
Physical stimulus – Lens power
Perceptual response – Clarity of the image

35. Why study psychophysics?
Psychophysical measurements are fundamental in clinical practice
Need to know the scientific basis for measuring vision
The results you get depend on the way you measure vision
New clinical tools will be developed after you graduate – you need the knowledge base to understand how they work and evaluate whether they are useful in your practice.
Psychophysics questions have been plentiful on the boards

37. Visual thresholds are the most common psychophysical measurement

41. Key in measuring thresholds: Try to keep all dimensions unchanged except the one being measured

45. There are many possible values of L,
But only 1 value (theoretically) for
threshold L
(demo)

51. For us to see, neural signals must leave the retina and travel to central brain structures.

52. In the early retinal cells (photoreceptors, bipolars, horizontal cells, most amacrines), there are only “graded potentials” (hyperpolarization and depolarization of the cell)
In order to send signals out of the retina, “action potentials” (“spikes”) must be generated and travel down the ganglion cell’s axon to the next location (lateral geniculate nucleus, then to visual cortex)

53. Graded potentials
The signal changes from graded potentials (voltage changes) into a “digital signal” in which the number of action potentials per second (firing rate) carries the visual signal.

54. We can “eavesdrop” on the neurons in the visual pathway with a microelectrode, nestled up against a neuron or its axon and record the responses (number of spikes per second) in response to visual stimuli.
B: Action potentials recorded from a single LGN neuron. The same stimulus (a spot of light positioned in the “receptive field” was presented many 20 times. A: a “histogram” of the cell’s responses

55. Action potentials recorded from a single LGN neuron
Neural fluctuations: the neuron sometimes responds more, sometimes less, to the same stimulus.
Also, the neuron has variable background (“maintained”) activity that makes it hard for the neuron to detect when the stimulus is present.

56. This leads us to consider threshold as a probability that a stimulus is detected and to find the stimulus value that is detected 50% of the time (or some other criterion value)

59. Figure 1-4. Idealized psychometric function for a threshold detection task using the Method of Constant Stimuli. The threshold stimulus value is obtained by drawing a horizontal line from the 50% value on the response axis to the psychometric function and then dropping a vertical line from the function to the test field intensity axis.

60. In Class Demo

61. Rule: Plot straight lines between data points

62. “Silliest Plotting Error”
Plot data points from left to right

63. “Silliest Plotting Error”
Plot data points from left to right

64. “Most Interesting Curves”

65. Figure 1-4. Idealized psychometric function for a threshold detection task using the Method of Constant Stimuli. The threshold stimulus value is obtained by drawing a horizontal line from the 50% value on the response axis to the psychometric function and then dropping a vertical line from the function to the test field intensity axis.

70. The Importance of Using Straight Lines to Connect Data Points
The data points are the only evidence we have of threshold
We assume a linear progression from one data point to the next
Can use linear interpolation to determine the threshold accurately

71. The Importance of Using Straight Lines to Connect Data Points
A dramatic example: If you measure vision incorrectly, you get an incorrect answer about how well a person sees.

72. Another way to mis-estimate threshold
We are looking for the 50% point, not the closest data point, so we use linear interpolation

73. We want to measure threshold as accurately as possible
We want to measure threshold as accurately as possible. Why be satisfied with “6” when 5.8 is more accurate?

76. What is another name for the psychometric function?
Threshold line
Frequency-of-seeing curve
Method of Constant Stimuli
Power function

78. In Class Demo

80. Developed during WWII to test bomb detonators

85. (Developed by Nobel Prize-winning auditory physiologist, Georg von Bekésy)

89. In Class Demo

90. Frequency with which LT is seen as equal to L

96. The catch trial gives the guessing rate
The catch trial gives the guessing rate. Then subtract the guessing rate from the data to get the “True percent of ‘Yes’ responses”

98. Do not memorize the formula! It isn’t used much.
Instead, people use the “forced choice” procedure

100. In Class Demo
Two-alternative forced choice

101. But if there are two alternatives (two-alternative forced-choice) you know the guessing rate is 0.5

102. Results from yesterday’s Method of Constant Stimuli Threshold Measurement
Figure 1-4. Idealized psychometric function for a threshold detection task using the Method of Constant Stimuli. The threshold stimulus value is obtained by drawing a horizontal line from the 50% value on the response axis to the psychometric function and then dropping a vertical line from the function to the test field intensity axis.

106. Results from yesterday’s Two-alternative Forced Choice Measurement

107. Intensity Discrimination Lab tomorrow
Groups C (1 – 3) and A (3 – 5)
List posted on bulletin board
Planning to have the lab unless UAB cancels classes due to snow
Snow amount is predicted to be small

108. 2009

110. Either way gives the same threshold, but it is easier to use a 75% threshold and not bother to correct

112. Principles of Psychophysical Measurement
Chapter 1
Principles of Psychophysical Measurement
Objectives:
Psychophysical Methods
Threshold
Constant Stimuli
Limits
Adjustment
Signal detection theory
Sensory Magnitude

114. Big point!!
Lesser point

115. Stimuli that are near threshold always are difficult to see
Stimuli that are near threshold always are difficult to see! Did I see that, or didn’t I?
The brain (comprised of neurons) must “decide” if a stimulus was present against a background of neural “noise”.

116. Your brain causes perception
Your brain causes perception. Cells in the brain do not respond to light. They respond because they are activated by a chain of cells that start with photoreceptors, which do “see” light.

117. We can “eavesdrop” on the neurons in the visual pathway with a microelectrode, nestled up against a neuron or its axon and record the responses (number of spikes per second) in response to visual stimuli.
B: Action potentials recorded from a single LGN neuron. The same stimulus (a spot of light positioned in the “receptive field” was presented many 20 times. A: a “histogram” of the cell’s responses

118. The visual system has to decide if a stimulus is present “on the fly” – as events happen
In studying how the visual system responds, we have the luxury of studying neural responses over many repeated trials
Use this information to understand why thresholds can be affected by “bias”

119. Signal + Noise
Noise

120. Below is a “peristimulus” histogram made from the responses to 30 stimulus repetitions like the three lines shown above.
We want to compare responses during
“noise” and “signal + Noise”

121. We are interested in how many action potentials are generated, over many stimulus presentations, during a 50 msec period when there is no stimulus (maintained discharge) and a 50 msec period when the stimulus is present.
Why 50 msec? Arbitrary, but it is about the amount of time the CNS seems to use.

122. Making a frequency distribution of neural responses during “noise” and “signal + noise”
50 ms “bins”
Stimulus + noise
“noise”
Number of action potentials in each bin 3 8 2 3
During “noise”, 0 spikes occur 1 time, 3 spikes occur 1 time, 2 occur 1 time 2 15
During “signal + noise”, 3 spikes occur 1 time, 8 spikes occur 1 time, 15 occur 1 time
Do this across 30 stimulus presentations to get a distribution of the frequency with which a certain number of spikes occurs

124. How can the brain “decide” if a near-threshold stimulus is present?
If a strong stimulus is presented, it produces many more action potentials during the “signal + noise” than are produced during the “noise”. But when a stimulus is near threshold, there is overlap between the number of spikes produced during “noise” and “signal + noise”

125. One can try various criteria –
Changing the criterion (the threshold one adopts) affects the pattern of hits, misses, false alarms and correct rejections
“The saga of the snake in the grass”
This changing threshold is partly responsible for fluctuations in threshold.

126. Imagine the situation faced by a mouse, needing to forage
for food, but worrying that a snake might be hanging around
and eat the mouse when the mouse goes out to eat

127. Set criterion low, to always detect the snake
If 6 or more action potentials, decide “snake!!”

128. Set criterion low, to always detect the snake
Problem: will also “see” snake some times when it is just the noise of the visual system
If 6 or more action potentials, decide “snake!!”

130. If set a low criterion (threshold), hit rate is perfect, but
Many false alarms

131. 10

133. So, try changing the criterion – mouse gets hungrier, willing to
So, try changing the criterion – mouse gets hungrier, willing to “take a chance”
If set a high criterion (threshold) have no false alarms but also fewer hits (more misses)

134. 10

135. Can calculate hit rate and false alarm rate for ANY criterion

136. Receiver Operating Characteristic (ROC) curve

139. The distributions on the previous slide would produce this
Hit Rate
Receiver Operating Characteristic
(ROC) Curve
1.0
0.8
0.6
0.4
0.2
0.0
0.0
0.2
0.4
0.6
0.8
1.0
False Alarm Rate

140. d’ (“d prime”) is a measure of the separation of two normal distributions.
d’ = the difference between the means of the “noise” and “signal plus noise” distributions divided by the common standard deviation of the two distributions.
d’ quantifies the detectability of the signal (small d’ = signal is hard to detect)

143. Big point: Where a neuron, or an entire creature (human or animal) sets its criterion depends on circumstances (fear vs. hunger which causes a change in “bias”). This contributes to threshold variability.

144. Use the “Payoff Game”

145. Use the “Payoff Game”

146. Skip the text on pages 24, 26, 27 and top of 28
“Signal detection theory can be used to control bias when measuring threshold”

147. Screening for refractive error:
Hits: Correct detection of refractive error
Correct rejection: pass the screening because child is emmetropic
False alarm (false positive): incorrectly refer for full exam based on screening (cost, concern, inconvenience)
Misses (false negative): fail to detect refractive error
Minimize false positives even though some refractive error is missed

148. Detecting ocular melanoma:
Hits: Correct detection of melanoma (refer for possible surgery)
Correct rejection: pass because no melanoma
False positive – incorrectly refer based on screening (alarm, cost, inconvenience)
Misses (false negatives): fail to detect melanoma (possible death)
Minimize false negatives even though some false positives occur

149. You will hear in clinic about the “sensitivity” and “specificity” of diagnostic techniques.
Sensitivity is the hit rate
Specificity is the absence of false alarms
So plot (1 – specificity) on an ROC curve
Want a diagnostic tool that has high sensitivity and high specificity

150. As was said the first day of class
Visual thresholds are the most common psychophysical measurement
“Do you see it?”

151. “What does it look like?”

152. Increased light intensity is “brighter” but how much brighter?
Increased spot size is “larger”, but how much larger?

165. “What does it look like?”

166. Principles of Psychophysical Measurement
Chapter 1
Principles of Psychophysical Measurement
Objectives:
Psychophysical Methods
Threshold
Constant Stimuli
Limits
Adjustment
Signal detection theory
Sensory Magnitude