De la síntesis de la Relación a la abundancia en la Historia

9am 5pm MLA LLA _{At each test time;}

y V y _{• subject tested to threshold in normal manner, then}

• subject repeat tested at same sites, but given a mix of tme and sham stimuli

In any sensory system under test, afferent signals perceived by the subject are a mix of the imposed test stimulus, and the background fluctuations inherent within the system. Therefore, the nearer the test signal is to the sensory threshold, the harder it is for the subject to distinguish between the stimulus and the ‘noise’ in the system; The possibility of spurious signals being reported as threshold sensation must be acknowledged. Also, the more willing the subject is to give a positive response to a threshold stimulus in the test situation, the more likely he is to guess in favour of a stimulus having been given.

The data presented so fer, were all based on the assunq)tion that the subject’s awareness of threshold touch, indicated by his verbal reports of perceived threshold sensations, was accurate and constant at all test times, and therefore gave a reliable indication of T-P threshold. It

disregarded the possibility that the incoming ‘noise’ could have been erroneously perceived, and reported, by the subject as the threshold stimulus. It also assumed that the subject’s willingness^

to report the test stimulus as being at sensory threshold, did not vary with time of day.

The data generated in experiment 3.1.4 was used to:

(i) check the accuracy of the normal method of threshold assessment, against a method in which the subject was challenged by a mix of true and blank stimuli, at moming and late aftemoon tests

(ii) test that the subject’s willingness to make a positive response was the same at both the moming and late aftemoon tests

21 young, healthy adults were tested to touch-pressure (T-P) threshold twice in the moming, using two methods of testing at the mid-sole of the feet, and repeat tested, using the two

methods, at the end of the aftemocm (Table 51). The tests were carried out during September and October.

(:) Comparison of two methods of T-P threshold assessment:

Subjects were tested to T-P threshold, in the moming and in the late aftemoon, and at each test time, the subject was assessed twice. The first moming test used the normal method of T-P threshold assessment. The second moming test delivered a mix of real and blank stimuh. Both tests were repeated again in the late aftemoon. The methodology is explained as a worked example of one subject’s results.

During the first test in the moming. real stimuli of ascending intensity were delivered to all points at the plantar surfece of the mid sole, in the manner described for all other T-P threshold tests. No wamings preceded the stimuli. The subject was asked to report all / any perceived sensations o f touch within mid sole during the test time.

Thresholds were determined, by the application of filaments in ascending order, starting with the lightest (2.24) filament. The value of the filament at which the subject reported a threshold touch sensation for each test point was recorded. Once threshold had been identified within the test point, it was not re-tested with a higher filament. This test was completed once all points within the test area had been assigned a touch threshold value.

TTie threshold for the first test was calculated as the mean of the log unit values at each test point

A total of 20 skin points were tested within the midsole. Under the conditions of the first test, the example subject reported a moming mean T-P threshold as 2.944 Log units (calculated as the mean of the values given in Table 52:

Table 52:: T-P thresholds at the midsole, m om ing first test = 2.944 log units

M LA LLA 2.68 2.9 3.12 3.34 2.68 3.12 3.12 3.12 3.12 2.68 2.68 2.9 3.12 2.9 2.68 2.9 3.34 2.68 2.9 2.9

During the second test in the moming. stimuli of ascending intensity were delivered to the same area of skin as in the first test, in the same manner, but allowing only 50% of the stimuli to make skin contact. The subject was told that each stimulus would be preceded by a verbal waming, but he was not told that some of the stimuli would be blanks. Each stimulus, whether tme or felse, was preceded by the same verbal warning, after which the subject was required to state whether he had detected a touch stimulus. The same range of aesthesiometer filaments used in the first test, were used in the second test, and were applied in the same order.

The order of application of the real and blank stimuli was the same for all filaments, all subjects, and all tests (Table 53). Real stimuli (T) were apphed to the same 10 points at the midsole and blank stimuli (F) to the remainder for each level of filament, regardless o f whether threshold had already been identified at that point. The subject reported filament contact, by stating ‘yes’ [y] or ‘no’ [n]. The range of possible subject responses to each test point, from each filament was: Ty; Fy; Fn; Tn. All responses to all (real or blank) filament contacts were recorded. The threshold for each test per subject was calculated, as the mean of his lowest Ty scores.

Table 53: Real (T) and B lank (F) Test Points at the M idsole

M LA LLA T T T F F F T T F F T F T T F F F T F T

Filaments were applied to all T points, starting with the lightest filament that the subject had identified in the first test. The subject’s response was recorded as a ‘y’ or ‘n’ response

RESPONSE To Filament 2.68 MLA LLA Ty Tn Tn Fn Fn Fn Tn Tn Fn Fn Tn Fn Tn Tn Fn Fn Fn Tn Fn _Ty RESPONSE To Filament 2.90 MLA LLA Ty Ty Ty Fn Fy Fy Ty Tn Fy Fn Tn Fn Ty Tn Fn _Fy Fy Ty Fy Ty RESPONSE To Filament 3.12 MLA LLA Ty Ty Ty Fn Fn Fn _Ty Tn Fn Fn _Ty Fn Tn _Ty Fn Fn Fn _Ty Fn _Ty RESPONSE To Filament 3.34 MLA LLA Ty _{Ty Ty} Fn Fn _{Fy Ty} Fy Fy Ty Fn Ty Ty Fn Fn Fy Ty Fn Ty

The T-P threshold for the second moming test was calculated as the mean o f the lowest Ty response from each test point, as shown in Table 22. In this example, the mean T-P threshold in the second moming test was 2.944 log units (So for this subject, by chance, both methods gave exactly the same moming T-P threshold value, although there was some variation in the raw data that was used to calculate the 2 mean values; see Tables 52 and 54)

Table 54: Second Test (morning) M ean Ty responses: = 2.944 log units.

MLA LLA 2.68 2.9 2.9 2.9 3.34 3.12 2.9 3.12 2.9 2.68

The same tests and calculations were repeated again, in the late aftemoon. Group means for each test method were compared, and are shown in Graph 23, and Table 55. The T-P thresholds calculated from the Ty responses alone were slightly lower (but not significantly lower - see Table 56 below) than those calculated by the ‘normal’ method, ie: the method used in the other experiments (Thresholds: ‘Normal’ method:- am: 3.40 +/- 0.08 log units; pm: 3.17 +/- 0.07 log

units; A am-pm: 0.23 +/- 0.06; True / False Method:- am: 3.28 +/- 0.09; pm: 3.10 +/- 0.07; A am- pm: 0.19+/- 0.05)

Graph 23: Comparison of T-P Thresholds, from the ‘Normal and ‘True / False’ Methods of Assessment

Comparison of T-P Thresholds, assessed by two methods

moming aftemoon moming

I

I

Ty mean. Ty mean Norm Norm aftemoon

The results of both first and both second tests were compared, using 2 factor analysis of variance (Table 56), where factor 1 = Test Method (first test; second test), and factor 2 = Test time (moming; late afternoon). This showed that there was a significant difference in T-P thresholds, due to time of test (p < 0.01), but no significant difference due to the two test methods (p > 0.2). There was no interaction between the variables (p > 0.7)

This experiment showed that thresholds, and diurnal variations of threshold, determined by the ‘normal’ method were not significantly different to those determined by a more rigorous methodology.

(ii) The Reliability of the Subjects’ C riteria, over Time

The same set of T-P threshold data was used to test the reliability, over time, of the subjects’ criterion of threshold, to show whether the observed diumal change was due to his being more willing to report sensation at the aftemoon test. This was done by comparing the Fy/Ty

responses from both test times (See also Appendix 1) for a subset of 12 subjects. The subset was made up of those subjects who had made marginal errors to filaments at or near threshold in both the moming and late aftemoon tests. The remaining 9 subjects who had not made errors

(subjects who did not score ‘Fy’ to any points) were excluded from this analysis. Their analysis is included in Appendix 1.

Fy and Ty responses reflect both the subject’s accuracy of response to the threshold stimulus, and his willingness to report a signal as a threshold stimulus. For example, if he lowers his threshold criterion, or if he is more willing to say 'yes’, he would raise his hit rate, but also

identify more imaginary stimuli (the false positive rate would rise). Alternatively, if he is more stringent in his threshold criteria, and is more inclined to say ‘no’ to a near-threshold stimulus, he will make less mistakes, but his overall detection rate will decline (the true positive rate would fall) (Sekuler et al, 1973).

The false positive and true positive (Fy ; Ty) responses from the data collected during the moming and late aftemoon second tests [see (i) above] were compared. NB; It was assumed that the both the false positive (the subject’s willingness to report the ‘noise’ in the system as a threshold sensation) and tme positive (the subject’s willingness to report the stimulus) responses, followed a normal distribution.

The subjects’ Fy and Ty responses, to the filaments to either side of their Ty-identified T-P threshold, for the moming and late afremoon second tests were averaged, and expressed as percentages. Using the criterion table (Table 57, see Levine and Parkinson, 1994) percentage Fy and Ty values were converted to criterion (z) values.

The same subject used in (i) as an example, is used again to illustrate this analysis. His results for the moming and late aftemoon second tests are shown in Tables 58 and 59.

Table 58: Example Subject: Mean Ty Response morning test,

together with his Fy and Ty responses to the filaments either side of the mean Ty response Mean Moming Ty responses (log units) Filaments to dther side of moming Ty T’hold Mean % Criterion (z) Table Value 2.90 3.12 2.944 Fy 0.60 0.00 0.30 30 0.52 Ty 0.70 0.80 0.75 75 -0.67

Table 59: Example Subject: Mean Ty Response Late Aftemoon Test,

together with his Fy and Ty responses to the filaments either side of the mean Ty response Mean Late

Aftemoon Ty responses (log units)

Filaments to either side of late aftemoon Ty T hold Mean % Criterion (z) Table Value 2.46 2.68 2.614 Fy 0.2 0.10 0.15 15 1.04 Ty 0.40 0.50 0.45 45 0.13

The same calculation was made for all 12 subjects (within the sangle of 21) who gave usefiil Fy values (the remaining subjects within the group did not make any errors of judgement, scoring

Fy = 0 to the filaments either side of the Ty mean threshold, either in the moming, the aftemoon, or both). The data for the remaining 9 subjects did not show a trend (Paired ‘t ’ = 0.679), and is shown in Appendix I.

The subject’s criterion value (Q for any test of signal detection can be calculated as the mean of his responses to the tme and the false stimuli, when these have been transformed to ‘z’ values (Table 43; See Appendix I, and Levine and Parkinson, 1994), so that:

C = 0 .5 [z „ + Z s n ]

where C is the subject criterion

Z n is the transformed felse positive (Fy) rate Z s n is the transformed tme positive (Ty) rate

The example subject showed a bias towards re^onding ‘yes’ in the moming (as the C value was negative) but had changed by the aftemoon to a positive C value, indicating that he was less likely to report the signal at that time (Diagram 30):

C.m=0.5 [ 0.52 + (_ 0.67)] = - 0.075 Cpm=0.5 [ 1.04+ ( 0 .1 3 ) ] = 0.585

When moming scores were compared with the late aftemoon scores, using a paired ‘t ’ test, it was seen that the subjects’ willingness to report threshold stimuli did not alter with the time

of day (p > 0.1, paired ‘t ’ test) (Table 60). If anything, subjects showed a slightly higher (though not of significance) criterion value at the aftemoon test, inferring that they tended to make less threshold guesses in the aftemoon.

These ejqseriments had shown that the oserved changes in T P threshold with time of day were unlikely to be due to a non-stringent method of data collection, nor due to the subject being more willing to respond to a touch stimulus, real or supposed, at the late afternoon test. The observed changes were more likely to be due to a diurnal variation in T-P

thresholds, showing as decreased thresholds by the end of the aftemoon, in comparison with moming values.

* * * * * * * * * *

Diagram 30: CRITERIA: where the dark area shows false alarm probabilities, with its left edge