March 23, 1888.] 



SCIENCE 



H5 



A Critique of Psychophysic Methods. 



I READ with care the comment by Dr. Boas upon my article in 

 ih& American Journal of Psyc/iology, 3J\A C3.xry away from it the 

 impression that there is less difference of opinion between us than 

 Dr. Boas supposes. The question is not one of fact, but of in- 

 terpretation. We all admit that there is a psychophysic fact for 

 which the word ' threshold ' is a good name ; but the important 

 question is, How shall we theoretically understand the conception, 

 and what place shall we allow it in the development of an experi- 

 mental psychology? Fechner makes it rank as by all means the 

 most important factor in psychophysics, and is willing to sacrifice 

 Weber's law before yielding the supreme and fundamental fact of 

 the threshold. He is led to this view by the method of the 'just 

 observable difference,' and by the neglect of the other two methods. 

 This entire structure I regard as reared upon an illogical basis, and 

 a psychophysics based upon the mathematical methods as very dif- 

 ferent and much sounder than the other. The threshold as a 

 practical, empirical fact, I not only fully admit, but even suggest 

 methods of further developing its utility ; but its theoretical impor- 

 tance with reference to the establishment of a psychophysic law I 

 regard as almost nil, its true importance lying in another direction. 

 This, I trust, defines my position clearly. A single illustration may 

 not be out of place. Dr. Boas says that a balance has a threshold, 

 and I accept the comparison. This threshold is something to be 

 eliminated, and that balance is the finest that has the least of this 

 characteristic. The theoretical balance upon which mechanics 

 works out its principles has no threshold. But apart from this, I 

 think the physicist will agree with me that it leads to more useful 

 and scientific conceptions to regard every particle that is placed 

 upon the pan of the balance as producing an effect alike in kind, 

 and differing only in degree from that produced by a mass suffi- 

 cient to turn the balance. There is no point where a new factor 

 enters, and the turning of the balance is a merely empirical fact. 

 Returning to the psychophysical methods, I should state the case 

 thus : it is generally admitted that the basis of the method of the 

 "right and wrong cases," as of the "average error," ultimately 

 rests upon the fact that the probabilities of my making errors of 

 various degrees follow the path traced by the probability curve. 

 This is the fundamental fact of the entire science of psychophysics. 

 Now, this curve is a continuous one, and has no break in it, no 

 point characterized by any special peculiarity, no threshold in any 

 true sense. 



A word as to my misrepresenting the views of my opponents. 

 The important point is, not what the upholders really do say, but 

 what logically follows from the position they take. If they do not 

 say what I attribute to them, it is because they are inconsistent; 

 and I have guarded myself against this misunderstanding by at 

 times stating, and elsewhere unmistakably implying, that I was 

 dealing with the logical consequences of the threshold theory, and 

 not with that particular portion of it that its adherents happened to 

 employ. 



The second point in Dr. Boas's criticism is a real difference of 

 opinion between us. He thinks " doubtful " answers should be 

 admitted in experimentation : I most emphatically object to them. 

 In my paper I regarded the objections to allowing such answers as 

 so necessarily following from the theory of the " right and wrong 

 cases " method, that a full statement of the reasons was super- 

 fluous. Any one of half a dozen reasons is enough to show the 

 impropriety of the " doubtful " answers. For instance : it is ad- 

 mitted that the methods should be as comparable, one with the 

 other, as possible. Now, the method of the " average error " 

 depending upon the same principle as that of the " right and wrong 

 cases," allows no doubtful answers. Again : there is no reason 

 for singling out " doubtful " answers as any thing peculiar. Why 

 not make a special rubric of unusually confident answers } And if 

 we do, as Dr. Boas suggests, make a threshold where doubtful 

 answers no longer occur, that threshold will vary so much in differ- 

 ent individuals, etc., that it will invalidate a large share of the 

 results. And what shall I say when some one else proposes a 

 threshold for another degree of confidence, say, the point where one 

 is sufficiently sure of the correctness of one's answer to risk money 

 upon it, and soon, ad infinitum ? If you mean that this subjective 

 feeling is worth taking account of, I fully concord, and will wel- 



come the skilful observation of this feeling as an important contri- 

 bution to psychophysics. Joseph Jastrow. 



Baltimore, March 12. 



On the Sense of Taste.' 



At the Philadelphia meeting of the American Association we 

 presented a paper upon the ' Delicacy of the Special Senses,' — a 

 topic upon which we have since continued our investigations from 

 time to time." 



The method pursued in the following experiments was as fol- 

 lows : — 



Solutions of known strength were made of the substances to be 

 lasted ; then, by successive dilutions, several series of solutions 

 were made from these, each one in the series being of one-half the 

 strength of the preceding one. The bottles containing these solu- 

 tions, and several bottles of water, were placed without regard to 

 order, and the person to be experimented upon was requested to 

 separate them into their proper groups by tasting them. In each 

 series the last solution was so dilute as to be beyond recognition. 

 All unrecognized solutions were classified as water. 



We chose for our tests the following typical substances. The 

 strength of the initial solution of each is given below. 



1. (Bitter) quinine, one part in 10,000 parts of water. 



2. (Sweet) cane-sugar, one part in 10 parts of water. 



3. (Acid) sulphuric acid, one part in 100 parts of water. 



4. (Alkaline) sodium bicarbonate, one part in ten parts of water. 



5. (Saline) sodium chloride, one part in 100 parts of water. 



The attempt was made to include other substances, as aromatics, 

 in the test ; but it was soon found that the odor betrayed their 

 presence without the aid of the sense of taste. 



Other investigators have added astringents as a sixth class, but 

 these substances are so often recognizable by odor, color, or some 

 special taste not purely astringent, that it was thought best not to 

 include them. 



Tests by the method above described were made upon 128 per- 

 sons ; 82 being male, and 46 female observers. 



The following table shows the amount of each substance which 

 could be detected by the average observer : — 



Substances. 



Male Observers detected. 



Female Observers detected. 



Quinine . . . 



I part i 



n 3g2,coo 



I part i 



n 456,000 



Sugar 



" ' 



199 





204 



Acid .... 



" ' 



2,080 





3,280 



Soda .... 



" ' 



98 





126 



Salt .... 





' 2,240 



" ' 



' i,q8o 



From the above results the following conclusions may be 

 drawn : — 



1. The sense of taste is vastly more delicate for bitter substances 

 than for any others. It is possible to detect quinine in a solution 

 that is only ^j'^jj the strength of a sugar solution, and we have pre- 

 viously shown {loc. cit.) that quinine is only ^ as bitter as strich- 

 nine. 



2. The order of delicacy is, bitter, acid, salt, sugar, and alkali. 



3. The sense of taste appears to be more delicate in women than- 

 in men. This is true in the case of all the substances excepting 

 salt. As we had found a similar difference in favor of female ob- 

 servers in an earlier and independent set of experiments, which 

 agreed in every essential particular with the results of the present 

 test, we do not regard it as an accidental difference, or as likely to 

 disappear in more extended investigations. 



Marked differences in the delicacy of the sense of taste of differ- 

 ent individuals were met with in the course of these experiments. 



1 Paper read at the New Yoric meeting of the American Association for the Ad- 

 vancement of Science, August, 1887. 



= See Relative Bitterness of Different Bitter Substances, by E. H. S. Bailey and 

 E. C. Franklin, in Proceedings o/the Kansas Academy 0/ Sciences, 1885 ; Relative 

 Sweetness of Sugars, by E. H. S. Bailey, in Report 0/ Kansas Board of Agriculture, 

 1884 ; The Sense of Smell, by E. L. Nichols and E. H. S. Bailey, in Nature, xxxv. 

 p. 74. 



