SDT primer

You have a sensor (1D continuous value).
You have to decide which is a signal and which
    is a noise, based on the sensor value.
When you classify the data as signal,
   you are aware of the signal.
SDT primer

You have a sensor (1D continuous value).
You have to decide which is a signal and which
    is a noise, based on the sensor value.
When you classify the data as signal,
   you are aware of the signal.


1) You collect samples.
2) You set the criteria for optimal discrimination.
3) You classify new data by comparing the
    sensor value and the criteria.
1) You collect samples.
1) You collect samples.
2) You set the criteria for optimal discrimination.
1) You collect samples (training data).
2) You set the criteria for optimal discrimination.
3) You classify new data by comparing the sensor value and
    the criteria.
3) You classify new data by comparing the sensor value and
    the criteria.


                           Criteria
3) You classify new data by comparing the sensor value and
    the criteria.


                           Criteria
3) You classify new data by comparing the sensor value and
    the criteria.


                           Criteria
3) You classify new data by comparing the sensor value and
    the criteria.


                           Criteria
3) You classify new data by comparing the sensor value and
    the criteria.


                           Criteria
3) You classify new data by comparing the sensor value and
    the criteria.


                           Criteria
3) You classify new data by comparing the sensor value and
    the criteria.


                           Criteria
3) You classify new data by comparing the sensor value and
    the criteria.


                           Criteria
3) You classify new data by comparing the sensor value and
    the criteria.


                           Criteria
3) You classify new data by comparing the sensor value and
    the criteria.


                           Criteria
3) You classify new data by comparing the sensor value and
    the criteria.


                           Criteria
3) You classify new data by comparing the sensor value and
    the criteria.


                           Criteria
3) You classify new data by comparing the sensor value and
    the criteria.


                           Criteria
3) You classify new data by comparing the sensor value and
    the criteria.


                           Criteria
SDT primer

1) You collect samples (training data).
2) You set the criteria for optimal discrimination.
3) You classify test data by comparing the sensor value and
    the criteria.

When you classify the data as signal,
  you are aware of the signal.

Let’s do it again, with different data set.
1) You collect samples (training data).
1) You collect samples (training data).
2) You set the criteria for optimal discrimination.
1) You collect samples.
2) You set the criteria for optimal discrimination.
3) You classify new data by comparing the sensor value and
    the criteria.
3) You classify new data by comparing the sensor value and
    the criteria.


                   Criteria
3) You classify new data by comparing the sensor value and
    the criteria.


                   Criteria
3) You classify new data by comparing the sensor value and
    the criteria.


                   Criteria
3) You classify new data by comparing the sensor value and
    the criteria.


                   Criteria
3) You classify new data by comparing the sensor value and
    the criteria.


                   Criteria
3) You classify new data by comparing the sensor value and
    the criteria.


                   Criteria
3) You classify new data by comparing the sensor value and
    the criteria.


                   Criteria
3) You classify new data by comparing the sensor value and
    the criteria.


                   Criteria
3) You classify new data by comparing the sensor value and
    the criteria.


                   Criteria
3) You classify new data by comparing the sensor value and
    the criteria.


                   Criteria
Recognition model (=> model-free)



        Unknown processes
                       classify
generate               with criteria (c=2)
                       awareness as decision


      Data (signal or noise)
Generative model (=> model-based)



      Processes with unknown parameters
      Noise: N(0,1); Signal: N(d’,1)

                           Estimate parameter
generate                   (d’ = 4) and
                           classify with criteria

      Data (signal or noise)
SDT primer



Processes with unknown parameters
Noise: N(0,1); Signal: N(d’,1)
SDT primer



     Processes with unknown parameters
     Noise: N(0,1); Signal: N(d’,1)

The sensitivity of the sensor is
    characterized as d’.
d’ is independent of criteria (c).
(The correct ratio depends on c.)
SDT primer



     Processes with unknown parameters
     Noise: N(0,1); Signal: N(d’,1)

The sensitivity of the sensor is
    characterized as d’.
d’ is independent of criteria (c).
(The correct ratio depends on c.)

OK, but we have no such sensor.
How to estimate d’ in psychophysics?
By changing criteria
By changing criteria


1) You set a criterion and classify samples.
2) You get the feedback (correct or incorrect).
3) You obtain data set 1 (with hit, miss, FA, CR).
4) Repeat 1)-3) with different criteria.
5) You reconstruct the distribution of samples.
6) You estimate d’.
1) You set a criterion and classify samples.
signal   noise
                                                  yes   hit ●     ○
                                                                 FA
                                                  no    miss ○ CR ●




1) You set a criterion and classify samples.
2) You get the feedback (correct or incorrect).
signal   noise
                                                  yes   hit ●     ○
                                                                 FA
                                                  no    miss ○ CR ●




1) You set a criterion and classify samples.
2) You get the feedback (correct or incorrect).
signal   noise
                                                  yes   hit ●     ○
                                                                 FA
                                                  no    miss ○ CR ●




1) You set a criterion and classify samples.
2) You get the feedback (correct or incorrect).
signal   noise
                                                  yes   hit ●     ○
                                                                 FA
                                                  no    miss ○ CR ●




1) You set a criterion and classify samples.
2) You get the feedback (correct or incorrect).
signal   noise
                                                  yes   hit ●     ○
                                                                 FA
                                                  no    miss ○ CR ●




1) You set a criterion and classify samples.
2) You get the feedback (correct or incorrect).
signal   noise
                                                  yes   hit ●     ○
                                                                 FA
                                                  no    miss ○ CR ●




1) You set a criterion and classify samples.
2) You get the feedback (correct or incorrect).
signal   noise
                                                  yes   hit ●     ○
                                                                 FA
                                                  no    miss ○ CR ●




1) You set a criterion and classify samples.
2) You get the feedback (correct or incorrect).
signal   noise
                                                  yes   hit ●     ○
                                                                 FA
                                                  no    miss ○ CR ●




1) You set a criterion and classify samples.
2) You get the feedback (correct or incorrect).
signal   noise
                                                  yes   hit ●     ○
                                                                 FA
                                                  no    miss ○ CR ●




1) You set a criterion and classify samples.
2) You get the feedback (correct or incorrect).
signal   noise
                                                  yes   hit ●     ○
                                                                 FA
                                                  no    miss ○ CR ●




1) You set a criterion and classify samples.
2) You get the feedback (correct or incorrect).
signal   noise
                                                  yes   hit ●     ○
                                                                 FA
                                                  no    miss ○ CR ●




1) You set a criterion and classify samples.
2) You get the feedback (correct or incorrect).
signal   noise
                                                  yes   hit ●     ○
                                                                 FA
                                                  no    miss ○ CR ●




1) You set a criterion and classify samples.
2) You get the feedback (correct or incorrect).
1) You set a criterion and classify samples.
2) You get the feedback (correct or incorrect).
3) You obtain data set 1 (with hit, miss, FA, CR).
4) Repeat 1)-3) with different criteria.
4) Repeat 1)-3) with different criteria.
4) Repeat 1)-3) with different criteria.
4) Repeat 1)-3) with different criteria.
4) Repeat 1)-3) with different criteria.
4) Repeat 1)-3) with different criteria.
4) Repeat 1)-3) with different criteria.
4) Repeat 1)-3) with different criteria.
5) You reconstruct the distribution of samples.
6) You estimate d’.
5) You reconstruct the distribution of samples.
6) You estimate d’.
5) You reconstruct the distribution of samples.
6) You estimate d’.
5) You reconstruct the distribution of samples.
6) You estimate d’.
5) You reconstruct the distribution of samples.
6) You estimate d’.
5) You reconstruct the distribution of samples.
6) You estimate d’.
5) You reconstruct the distribution of samples.
6) You estimate d’.
How do you change the criteria?
How do you change the criteria?
1) Confidence rating (Human study)
How do you change the criteria?
1) Confidence rating (Human study)
How do you change the criteria?
1) Confidence rating (Human study)
                 No      Yes
How do you change the criteria?
1) Confidence rating (Human study)
  Very sure   Sure   Uncertain Uncertain   Sure   Very sure
How do you change the criteria?
1) Confidence rating (Human study)
  Very sure   Sure   Uncertain Uncertain   Sure   Very sure




2) By changing value or probability (animal study)
How do you change the criteria?
1) Confidence rating (Human study)
  Very sure   Sure   Uncertain Uncertain   Sure   Very sure




2) By changing value or probability (animal study)
How do you change the criteria?
1) Confidence rating (Human study)
  Very sure   Sure   Uncertain Uncertain   Sure   Very sure




2) By changing value or probability (animal study)
How do you change the criteria?
1) Confidence rating (Human study)
  Very sure   Sure   Uncertain Uncertain   Sure   Very sure




2) By changing value or probability (animal study)
How do you change the criteria?
1) Confidence rating (Human study)
  Very sure   Sure   Uncertain Uncertain   Sure   Very sure




2) By changing value or probability (animal study)

信号検出理論の解説 (Signal detection theory, a primer)

  • 1.
    SDT primer You havea sensor (1D continuous value). You have to decide which is a signal and which is a noise, based on the sensor value. When you classify the data as signal, you are aware of the signal.
  • 2.
    SDT primer You havea sensor (1D continuous value). You have to decide which is a signal and which is a noise, based on the sensor value. When you classify the data as signal, you are aware of the signal. 1) You collect samples. 2) You set the criteria for optimal discrimination. 3) You classify new data by comparing the sensor value and the criteria.
  • 3.
  • 4.
    1) You collectsamples. 2) You set the criteria for optimal discrimination.
  • 24.
    1) You collectsamples (training data). 2) You set the criteria for optimal discrimination. 3) You classify new data by comparing the sensor value and the criteria.
  • 25.
    3) You classifynew data by comparing the sensor value and the criteria. Criteria
  • 26.
    3) You classifynew data by comparing the sensor value and the criteria. Criteria
  • 27.
    3) You classifynew data by comparing the sensor value and the criteria. Criteria
  • 28.
    3) You classifynew data by comparing the sensor value and the criteria. Criteria
  • 29.
    3) You classifynew data by comparing the sensor value and the criteria. Criteria
  • 30.
    3) You classifynew data by comparing the sensor value and the criteria. Criteria
  • 31.
    3) You classifynew data by comparing the sensor value and the criteria. Criteria
  • 32.
    3) You classifynew data by comparing the sensor value and the criteria. Criteria
  • 33.
    3) You classifynew data by comparing the sensor value and the criteria. Criteria
  • 34.
    3) You classifynew data by comparing the sensor value and the criteria. Criteria
  • 35.
    3) You classifynew data by comparing the sensor value and the criteria. Criteria
  • 36.
    3) You classifynew data by comparing the sensor value and the criteria. Criteria
  • 37.
    3) You classifynew data by comparing the sensor value and the criteria. Criteria
  • 38.
    3) You classifynew data by comparing the sensor value and the criteria. Criteria
  • 39.
    SDT primer 1) Youcollect samples (training data). 2) You set the criteria for optimal discrimination. 3) You classify test data by comparing the sensor value and the criteria. When you classify the data as signal, you are aware of the signal. Let’s do it again, with different data set.
  • 40.
    1) You collectsamples (training data).
  • 41.
    1) You collectsamples (training data). 2) You set the criteria for optimal discrimination.
  • 55.
    1) You collectsamples. 2) You set the criteria for optimal discrimination. 3) You classify new data by comparing the sensor value and the criteria.
  • 56.
    3) You classifynew data by comparing the sensor value and the criteria. Criteria
  • 57.
    3) You classifynew data by comparing the sensor value and the criteria. Criteria
  • 58.
    3) You classifynew data by comparing the sensor value and the criteria. Criteria
  • 59.
    3) You classifynew data by comparing the sensor value and the criteria. Criteria
  • 60.
    3) You classifynew data by comparing the sensor value and the criteria. Criteria
  • 61.
    3) You classifynew data by comparing the sensor value and the criteria. Criteria
  • 62.
    3) You classifynew data by comparing the sensor value and the criteria. Criteria
  • 63.
    3) You classifynew data by comparing the sensor value and the criteria. Criteria
  • 64.
    3) You classifynew data by comparing the sensor value and the criteria. Criteria
  • 65.
    3) You classifynew data by comparing the sensor value and the criteria. Criteria
  • 66.
    Recognition model (=>model-free) Unknown processes classify generate with criteria (c=2) awareness as decision Data (signal or noise)
  • 67.
    Generative model (=>model-based) Processes with unknown parameters Noise: N(0,1); Signal: N(d’,1) Estimate parameter generate (d’ = 4) and classify with criteria Data (signal or noise)
  • 68.
    SDT primer Processes withunknown parameters Noise: N(0,1); Signal: N(d’,1)
  • 69.
    SDT primer Processes with unknown parameters Noise: N(0,1); Signal: N(d’,1) The sensitivity of the sensor is characterized as d’. d’ is independent of criteria (c). (The correct ratio depends on c.)
  • 70.
    SDT primer Processes with unknown parameters Noise: N(0,1); Signal: N(d’,1) The sensitivity of the sensor is characterized as d’. d’ is independent of criteria (c). (The correct ratio depends on c.) OK, but we have no such sensor. How to estimate d’ in psychophysics?
  • 71.
  • 72.
    By changing criteria 1)You set a criterion and classify samples. 2) You get the feedback (correct or incorrect). 3) You obtain data set 1 (with hit, miss, FA, CR). 4) Repeat 1)-3) with different criteria. 5) You reconstruct the distribution of samples. 6) You estimate d’.
  • 73.
    1) You seta criterion and classify samples.
  • 74.
    signal noise yes hit ● ○ FA no miss ○ CR ● 1) You set a criterion and classify samples. 2) You get the feedback (correct or incorrect).
  • 75.
    signal noise yes hit ● ○ FA no miss ○ CR ● 1) You set a criterion and classify samples. 2) You get the feedback (correct or incorrect).
  • 76.
    signal noise yes hit ● ○ FA no miss ○ CR ● 1) You set a criterion and classify samples. 2) You get the feedback (correct or incorrect).
  • 77.
    signal noise yes hit ● ○ FA no miss ○ CR ● 1) You set a criterion and classify samples. 2) You get the feedback (correct or incorrect).
  • 78.
    signal noise yes hit ● ○ FA no miss ○ CR ● 1) You set a criterion and classify samples. 2) You get the feedback (correct or incorrect).
  • 79.
    signal noise yes hit ● ○ FA no miss ○ CR ● 1) You set a criterion and classify samples. 2) You get the feedback (correct or incorrect).
  • 80.
    signal noise yes hit ● ○ FA no miss ○ CR ● 1) You set a criterion and classify samples. 2) You get the feedback (correct or incorrect).
  • 81.
    signal noise yes hit ● ○ FA no miss ○ CR ● 1) You set a criterion and classify samples. 2) You get the feedback (correct or incorrect).
  • 82.
    signal noise yes hit ● ○ FA no miss ○ CR ● 1) You set a criterion and classify samples. 2) You get the feedback (correct or incorrect).
  • 83.
    signal noise yes hit ● ○ FA no miss ○ CR ● 1) You set a criterion and classify samples. 2) You get the feedback (correct or incorrect).
  • 84.
    signal noise yes hit ● ○ FA no miss ○ CR ● 1) You set a criterion and classify samples. 2) You get the feedback (correct or incorrect).
  • 85.
    signal noise yes hit ● ○ FA no miss ○ CR ● 1) You set a criterion and classify samples. 2) You get the feedback (correct or incorrect).
  • 86.
    1) You seta criterion and classify samples. 2) You get the feedback (correct or incorrect). 3) You obtain data set 1 (with hit, miss, FA, CR).
  • 87.
    4) Repeat 1)-3)with different criteria.
  • 88.
    4) Repeat 1)-3)with different criteria.
  • 89.
    4) Repeat 1)-3)with different criteria.
  • 90.
    4) Repeat 1)-3)with different criteria.
  • 91.
    4) Repeat 1)-3)with different criteria.
  • 92.
    4) Repeat 1)-3)with different criteria.
  • 93.
    4) Repeat 1)-3)with different criteria.
  • 94.
    4) Repeat 1)-3)with different criteria.
  • 95.
    5) You reconstructthe distribution of samples. 6) You estimate d’.
  • 96.
    5) You reconstructthe distribution of samples. 6) You estimate d’.
  • 97.
    5) You reconstructthe distribution of samples. 6) You estimate d’.
  • 98.
    5) You reconstructthe distribution of samples. 6) You estimate d’.
  • 99.
    5) You reconstructthe distribution of samples. 6) You estimate d’.
  • 100.
    5) You reconstructthe distribution of samples. 6) You estimate d’.
  • 101.
    5) You reconstructthe distribution of samples. 6) You estimate d’.
  • 102.
    How do youchange the criteria?
  • 103.
    How do youchange the criteria? 1) Confidence rating (Human study)
  • 104.
    How do youchange the criteria? 1) Confidence rating (Human study)
  • 105.
    How do youchange the criteria? 1) Confidence rating (Human study) No Yes
  • 106.
    How do youchange the criteria? 1) Confidence rating (Human study) Very sure Sure Uncertain Uncertain Sure Very sure
  • 107.
    How do youchange the criteria? 1) Confidence rating (Human study) Very sure Sure Uncertain Uncertain Sure Very sure 2) By changing value or probability (animal study)
  • 108.
    How do youchange the criteria? 1) Confidence rating (Human study) Very sure Sure Uncertain Uncertain Sure Very sure 2) By changing value or probability (animal study)
  • 109.
    How do youchange the criteria? 1) Confidence rating (Human study) Very sure Sure Uncertain Uncertain Sure Very sure 2) By changing value or probability (animal study)
  • 110.
    How do youchange the criteria? 1) Confidence rating (Human study) Very sure Sure Uncertain Uncertain Sure Very sure 2) By changing value or probability (animal study)
  • 111.
    How do youchange the criteria? 1) Confidence rating (Human study) Very sure Sure Uncertain Uncertain Sure Very sure 2) By changing value or probability (animal study)