112 PSYCHOLOGY. 



ideas are required. Most of those who have upheld the 

 thesis in question have, however, made a more complex, 

 supposition. What they have meant is that perception is 

 a mediate inference, and that the middle term is unconscious. 

 When the sensation which I have called ' this ' (p. 83, supra) 

 is felt, they think that some process like the following runs 

 through the mind : 



' This ' is M ; 



but M is A ; 



therefore ' this ' is A.* 



Now there seem no good grounds for supposing this, 

 additional wheelwork in the mind. The classification of 

 * this ' as M is itself an act of perception, and should, if all 

 perception were inference, require a still earlier syllogism for 

 its performance, and so backwards in infinitum. The only 

 extrication from this coil would be to represent the process 

 in altered guise, thus : 



'This' is like those; 

 Those are A ; 

 Therefore * this ' is A. 



The major premise here involves no association by conti- 

 guity, no naming of those as M, but only a suggestion of 

 unnamed similar images, a recall of analogous past sensa- 

 tions with which the characters that make up A were habit- 

 ually conjoined. But hers again, what grounds of fact are 

 there for admitting this recall ? We are quite unconscious 

 of any such images of the past. And the conception of all 

 the forms of association as resultants of the elementary fact 

 of habit-worn paths in the brain makes such images entirely 

 superfluous for explaining the phenomena in point. Since 

 the brain-process of ' this,' the sign of A, has repeatedly 

 been aroused in company with the process of the full object 

 A, direct paths of irradiation from the one to the other must 

 be already established. And although roundabout paths 

 may also be possible, as from 'this' to 'those, and then 



* When not all M, but only some M, is A, when, in other words, M is 

 ' undistributed ' the conclusion is liable to error. Illusions would thus be 

 logical fallacies, if true porceptions were valid syllogisms. They would 

 draw false conclusions from undistributed middle terms. 



