522 REPORTS ON THE STATE OF SCIENCE, ETC. 



any stimulus, experienced by itself, against the background of memorised 

 experiences of other stimuli of the same kind will be the same as we would 

 assign to it in a series of stimuli experienced simultaneously. If we mis- 

 takenly estimate sensation intensities on the unjustifiable assumption that 

 the sense distances in a so-called mean-gradation series are equal, we are 

 led to the mist^en conclusion that sensation intensity tends on the whole 

 to increase by equal amounts for equal increments of the ratio of stimulus 

 intensity (or that S varies as log /) ; then when we experience some stimulus 

 by itself, as when we observe a single bright object or hear a single sound, 

 we tend to estimate the sensation intensity on the same false (or rather 

 meaningless) scale by comparison with memories of previous experiences 

 of the sensation intensities corresponding to known stimulus intensities. 

 Actually we are not comparing sensation intensities at all, but are attempting 

 to place the stimulus in a series of its memorised predecessors, graded in 

 terms of the recognitive relation of ' sameness ' of relative intensity ratios. 



It cannot be over-emphasised, because it is so consistently overlooked, 

 that when we perceive a light or a sound or any other perceptible thing our 

 resulting impression, and the judgment we base on it is not of our sensations 

 but of those features of the environment from which we receive the stimulus. 

 That is the normal function of perception : our judgment is about the 

 objective intensity of the stimulus, and if we say of the members of a group 

 of stimuli graded by the mean-gradation criterion that the intensities 

 ' appear ' to differ by equal amounts, when in fact they do not differ by equal 

 amounts, we are merely asserting that our senses are misleading us and 

 providing wrong information about the phenomena observed. But our 

 senses do not grossly mislead us (except in unfamiliar circumstances). In 

 general they inform us reasonably correctly of the relations exhibited by the 

 phenomena : that is what they have been evolved for. It is only if we 

 misinterpret these relations that we are led to false conclusions. 



So, when we are trying to estimate the loudness of a sound, for example, 

 it is not the subjective intensity of our sensation we are estimating on any 

 scale, true or false, but the objective intensity of the sound. Whatever the 

 actual relation may be between sound intensity and sensation intensity it 

 merely serves the purpose of leading to an intuitive judgment about the 

 objective sound, and if we so interpret this judgment as to grade the sound 

 on any scale of magnitude but the right one — the scale of stimulus intensi- 

 ties — we are not discovering any fact about sensation intensities, but are 

 simply making a mistake about the objective intensities. Thus the pre- 

 valent idea that the sensations of brightness, loudness, etc., vary approxi- 

 mately as the logarithm of the stimulus intensity is devoid of any basis 

 and is neither proved by, nor even suggested by, mean-gradation 

 experiments. 



All these experiments can tell us is whether, on the whole, the operation 

 of our sensory system is such as to provide us with reasonably accurate 

 information about the relation structure of the objective world. If any 

 single act of perception provided us with absolutely accurate information 

 about the relations perceived, all mean-gradation series would consist of 

 intensities each in the same ratio to the next for the reasons we have dis- 

 cussed. But we would scarcely expect any act of perception to provide 

 absolutely accurate information about objective relation structure. As we 

 have already remarked, what we mean by the true objective relation structure 

 is a synthesis of an enormous number of relations observed in different 

 ways at different times. For any individual act of perception to provide 

 information absolutely consistent with this synthetic whole would require 

 a uniqueness of relation between stimulus and response only obtainable 



