84 ELIZABETH LOCKWOOD THOMPSON 



the formation of a weak association between the two stimuU 



used. Figs. 23-28, Plate VIII. 



15. The last series of experiments (paragraph 13) shows 



results comparable to those obtained in the first part of the 



present work which indicated an association formed between 



two dissimilar stimuli, food and pressure, in a series of tests 



based on the Pawlow salivary method. It does not show ability 



to learn the labyrinth. 



8. Discussion 



The foregoing attempt to learn whether the snail can solve 

 a simple vertical U- or Y-shaped labyrinth falls into three series. 



1. In the first of these there was used a U-shaped labyrinth 

 supported on a stem and bordered by a paling of wires so that 

 the snails could crawl up but one side of it. At the top of the 

 stem they had a choice of a right or left path. If the right 

 path were chosen it led to the right vertical bar of the U from 

 the tO}) of which the snail could not reach the surface of the 

 water. If the left path were chosen it led to the vertical bar 

 of the U, longer than the right, from the top of which the snail 

 could fill its lung with air. The upward course of the snail 

 is in search of air. Choice of the right path resulted in failure 

 to get air which was assumed to act as punishment. Such a 

 choice was regarded as incorrect. Choice of the left path led 

 to the attainment of air, which was regarded as reward and 

 such choice was registered as correct. 



Preliminary tests showed that the snails have a slight pref- 

 erence for the right path, a preference possibly related to the 

 asymm.etry of the animal. It was to be expected that this right 

 preference would appear at the beginning of the series of learning 

 tests, but that it would be soon corrected by the result of the 

 learning process. If thd snails actually learned and a curve 

 were plotted showing their percentage of errors from each succes- 

 'sive day's experience, this curve would start somewhere above 

 the 50% level on account of the right i^reference, which would 

 result in an excess of incorrect choices. The curve would then 

 gradually descend, through learning, toward the zero level with 

 decrease in error present. If the snail did not learn and were 

 in no way affected by its experiences with the labyrinth the 

 ■curve of errors might be expected to follow the 50% level or 

 somewhat above it, while fluctuating above and below. 



