QUANTITATIVE ESTIMATES OF SENSORY EVENTS 323 



from a mechanism of absolutely constant properties, deliberately designed 

 and constructed for the purpose. We do not look for such machine-like 

 behaviour from sensory mechanisms, made of living material necessarily 

 influenced by changes in our bodily conditions, and not designed, in the 

 usual sense of the term, but simply developed by the gradual processes of 

 evolution to become more and more serviceable to us. Evolution is a 

 gradual process. Our sensory mechanisms are doubtless much better suited 

 for providing us with the kind of information w^e want from them than 

 were those of our early forefathers, but are just as surely less well suited 

 for that purpose than they might be, and presumably may be some day. 



Although, as I have tried to emphasise, we do not consciously judge our 

 sensation intensities in an act of perception but judge the objective intensi- 

 ties, the sensation intensities are nevertheless the only basis of the judgment ; 

 and there must be a relation-structure in the psychological content of the 

 sense impression corresponding to the relation structure of the phenomena 

 perceived. We cannot expect from the nature of our sensory machinery 

 that exactly similar psychological relation structures will be produced by the 

 same phenomenal relation structure on every occasion and in all conditions 

 of observation. Repeated experience of the observation of any permanent 

 phenomenal relation structure will, however, establish a norm among the 

 various psychological structures produced by it at different times, and this 

 norjti will be the psychological structure which produces in us the reaction 

 which we interpret as perceiving the objective structure. Now if on any 

 individual occasion this same objective structure produces a psychological 

 structure different from the norm we shall ' perceive ' not the actual pheno- 

 menal structure, but another one for which the actual psychological structure 

 is itself the associated norm. In other words our judgment of the external 

 objects will be in error. Our judgments would, in fact, very frequently 

 be in error from this cause if they depended entirely on the immediate sense 

 impression ; but in the ordinary observation of the world our mistaken 

 impressions are corrected by another effect of association, known to 

 psychologists as regression towards the phenomenon. When we receive a 

 sense impression from a familiar object (or from one which we assume, 

 correctly or otherwise, to be a familiar object) which is not quite consistent 

 with the phenomenal structure that we normally associate with the object, 

 associative reflexes originating in accumulated experience of the relation 

 structure which ought to be perceived, come into operation and correct our 

 impression either to what it ought to be or to something much nearer this 

 than it would be in the absence of such associative reflexes. We are not 

 concerned here with either the physiological or psychological machinery of 

 phenomenal regression, but merely have to note that it is one of the most 

 important agencies in preserving the correspondence between our judgments 

 of phenomenal relation structures and the structures themselves despite a 

 somewhat imperfect correspondence between our immediate sensory re- 

 actions and the objective conditions which evoke them. 



Now it is evident that the effectiveness of this corrective agency — 

 phenomenal regression — since it depends on correlative associations, will be 

 most strongly developed for the observation of familiar phenomenal group- 

 ings encountered regularly in everyday life. Further, it cannot work 

 miracles. Even a familiar phenomenal grouping, if observed under con- 

 ditions for which the immediate sensory impression differs too much from 

 the norm associated with it, will be misjudged and will seem different from 

 what it really is. 



From its nature, phenomenal regression will be inoperative in circum- 

 stances which do not call up the correlative associations on which it depends. 



