PERCEPTION 



1613 



A/ 



XO"»Y 



FIG. 10. Figures used in testing form perception in honey 

 bees. The bees could be trained to distinguish each of the 

 figures in the upper row from each in the lower row, but failed 

 to distinguish among those in the upper row, or among any of 

 those in the lower row. [Modified from Hertz (J07).] 



analogous difficulties with mirror images; thus, they 

 failed to learn a distinction between \ and /. (How- 

 ever, Rudel's finding that C and 3 are practically 

 indistinguishable for small children, while u and n 

 are easily discriminated, does not seem to conform to 

 Sutherland's hypothetical scanning mechanism, nor 

 would ii seem to be predictable from an earlier theor) 

 of shape recognition advanced by Dcutsch (103) 

 (More recent observations by Sutherland himself 

 indicate that for octopus, too, there is much less diffi- 

 culty with u and n than with C and 3, although 

 the difference between the two tasks is not as great 

 as in the child. ) 



An incidental outcome of Sutherland's work is a 

 new and more convincing interpretation of Boycott & 

 Young's well-known report of altered memory for 

 visual form in the octopus following extirpation of the 

 vertical lobe system from the animal's brain (58). 

 Such animals tend to lose discrimination, and attack 

 both positive and negative shapes. 1 - However, some of 

 the animals can be retrained by presenting shapes at 

 very short intervals (a situation which interferes with 

 learning in normal animals, since they stop attacking 

 shapes to which they are frequently exposed). 



Young and Boycott have interpreted these observa- 

 tions by assuming that the negative memory in ani- 

 mals with verticalis lesions becomes very short-lived, 

 and that it can be maintained only by frequent expo- 

 sures (every 5 to 10 mini to the negative shape, even 



'- Although this effect is often quoted as being specific for 

 visual perception, it exists just as well for tactile discrimination 

 in a blinded octopus lacking the vertical lobe (533). 



if the shape is not attacked. Since Sutherland (460) 

 has shown that "exposure to a shape tends to reduce 

 the tendency to attack a shape" in the normal octopus, 

 it is likely that '"verticalis removal may raise general 

 tendency to attack and that discrimination can only- 

 show up when this tendency is reduced, and that this 

 reduction may be brought about by very frequent 

 presentation of shapes." 



salticidae. The octopus and other cephalopods are 

 not the only invertebrates whose eyes are capable of 

 forming single images. The jumping spiders 

 (Salticidae) possess such eyes, and apparently depend 

 on them in courtship and in their recognition of prev 

 (190, 221, 222). Heil's Held studies (190) describe 

 visual recognition of dead (and hence, immobile 1 

 prey. Such an achievement is unexpected, since it has 

 been assumed that most invertebrates require mining 

 targets for adequate perception. Laboratory studies 

 similar to those in the octopus would be desirable. 



INVERTEBRATES Willi COMPOUND EYES. By far the IllOSt 



pressing need, however, is for further elucidation of 

 pattern vision in higher invertebrates with compound 

 eyes. That visual orientation is important in bees has 

 been made clear b\ von Frisch (499, 500) but the wa\ 

 in which visual patterns are recognized by the bees 

 remains elusive Hertz, in numerous studies (207- 

 2119), demonstrated that bees distinguish star-shaped 

 from closed flower patterns, with preference for those 

 with radiating (and hence, with broken) contours 

 (see tig. 101. Zerrahn (557 1 showed further that the 

 bees' preference increased directly with increasing 

 numbers of contours within a given test pattern. 

 Wolf (546) found that rotating patterns were chosen 

 in preference to identical stationary ones; flickering 

 sources attracted more bees in direct proportion to the 

 rate of flicker. All these observations seemed to suggest 

 that "pattern" might reduce to "flicker,' i.e. the num- 

 ber of ommatidia stimulated in rapid succession as the 

 bee flies over a pattern rich in black-white contrast. 



This cannot be the whole story, however, since 

 von Frisch had shown as early as 191 4 [see also Knoll 

 (264) and Friedlander (133)] that bees can be trained 

 to distinguish bipartite disks (e.g. one lateral half 

 blue, the other half yellow), depending on the right- 

 left pattern (e.g. yellow-blue vs. blue-yellow). Such an 

 accomplishment in an animal with compound eyes is 

 all the more remarkable if one recalls the inability of 

 the octopus to distinguish mirror-image patterns. 

 Even more puzzling are the demonstrations by Hertz 

 (210) that with some patterns, bees can be trained to 



