SENSORY CAPACITIES AND INTELLIGENCE OF DOGS 



few years ago, in Berlin. These are 

 reported by Craig (7), who, however, 

 fails to give the original reference. The 

 tests were open not only to police dogs 

 but to all comers, and retests were per- 

 mitted after failures when so desired. A 

 number of prize winning dogs of various 

 breeds were entered and tested individu- 

 ally. Four types of test were given. The 

 first involved the following of a fresh 

 human trail among other recently made 

 trails. No dog proved able to do this 

 reliably, although most were able to 

 follow a fresh trail over older trails. 

 Among fresh trails the dogs often switched 

 from one to another, the angles of inter- 

 section seeming to play a prominent part 

 in determining which trail was to be 

 followed. The tendency was for the dog 

 to follow any fresh trail extending in the 

 same general direction taken by the trail 

 he had just been following, i.e. if the 

 trail he was following in a northerly 

 direction turned abruptly east, at a point 

 where it intersected a second fresh trail 

 which from that point extended northerly, 

 the dog usually followed the second trail. 



The second test involved the following 

 of an old trail over ground entirely free 

 from other trails. Twenty tests over 

 trails from 5^ to 6| hours old resulted in 

 absolute failure. Trails were followed 

 well only when not more than 30 minutes 

 old. 



The third test involved the selection 

 from a group of objects of the one pre- 

 viously handled by a person who had been 

 pointed out to the dog. In other words, 

 this corresponds to Lohner's experiment 

 except that other individuals than the 

 dog's master were used. In no case did a 

 dog make a correct choice more often than 

 he might have by chance. 



The last test was the converse of the 

 previous one. The dog was given the 

 glove of a person and told to select the 



owner, who, together with nine other 

 men, stood in line with their backs to the 

 dog. In no case did the men know whose 

 glove had been given the dog to smell. 

 This precaution prevented the "guilty" 

 man from revealing his identity in any 

 way. The dogs all failed in this test as 

 completely as in the previous ones. 



The conclusions drawn were that those 

 cases of successful detection of criminals 

 by dogs (as demonstrated by subsequent 

 confessions) were probably made on the 

 basis of slight movements of the criminal 

 resulting from his fear of being detected. 

 Those cases where a dog had apparently 

 trailed a criminal to his lair were inter- 

 preted as examples of the Clever Hans 

 error. It was supposed that the police 

 sergeant holding the leash on which the 

 dog was led while trailing often had a 

 sagacious guess as to the probable hiding 

 place of the criminal and, all unwittingly, 

 led the dog to that place, rather than 

 being led by the dog. In other words it 

 was a case similar to the "muscle-reading" 

 by which so-called mind-readers often 

 successfully entertain parlor audiences. 

 As a result of such tests, so Craig reports, 

 the use of dogs in criminal detection is 

 now forbidden by the Prussian government. 

 How are we to interpret these results? 

 It seems to the writers that these negative 

 results cannot be taken as definitely 

 proving a lower degree of olfactory acuity 

 in dogs than had previously been granted 

 them. It is more probable, in view of the 

 other evidence in favor of extraordinary 

 acuity, that these tests indicate that the 

 problems were not properly set for the 

 dogs. To use, for the sake of simplicity, 

 an anthropomorphic term, we have no 

 evidence that the dogs understood what 

 was expected of them. Negative conclu- 

 sions are always difficult to support, and 

 they are certainly not justified in the 

 absence of evidence that the problem at 



