References and Notes measurement of nine morphological characters uted between 1800 and 3700 hertz. The in 20 male and female E. coqui revealed a prolow- and mid-frequency units probably I. D. F. Austin and A. Schwartz, Copeia 1, 188 sexual dimorphism in size of all nine nounced (1975); A. Schwartz and R. Thomas, A Checkcharacters; for example, mean snout-to-vent derive from the amphibian papilla, List of West Indian Amphibians and Reptiles length: male, 37.2 mm and female, 52.2 mm; (Carnegie Museum of Natural History, Pittswhereas the high-frequency units origimean longitudinal tympanum diameter: male, 1975). 2.0 mm and female, 2.9 mm [R. Thomas, Q. J. nate in the basilar papilla (11). The histo- 2. burgh, P. M. Narins and R. R. Capranica, in preparaFla. Acad. Sci. 28, 375 (1966)]. The shift in fregram reveals a large population of fibers 3. tion. quency sensitivity between males and females N. B. Reyes Campos, Caribb. J. Sci. 11, 209 might be related to sexual dimorphism, such as a with BEF's in the range of the "Co" (1971). size difference in the auditory organs of the two durations, internote intervals, and repsexes. In addition, the characteristics of the note, whereas there are few units with 4. Call-note etition rates used in acoustic stimuli for male mating call of male E. coqui vary with the elevaplayback experiments were chosen from recordBEF's in the range of the "Qui" note. tion within the rain forest (G. E. Drewry, personof males from the same ings of typical calls al communication; P. M. Narins and R. R. CapNeither the low-frequency nor the highpopulation. These were recorded and supplied ranica, unpublished data). Furthermore, males at by G. E. Drewry. frequency populations in this histogram higher elevations are larger and produce lower5. L. S. Frishkopf, R. R. Capranica, M. H. Goldcalls than those at lower altitudes. This pitched coincide with the energy in the male's stein, Jr., Proc. IEEE 56, 969 (1968); A. J. M. raises the interesting question of whether the Moffat and R. R. Capranica, J. Acoust. Soc. call. Low-frequency sensitivity may play frequency sensitivity of the inner ear in an aniAm. 55, 480 (1974). mal of this species varies with its body size. a role in detection of predators for some 6. R. R. Capranica and A. J. M. Moffat, J. Comp. 14. The Q values for the high-frequency fibers in 100, 231 (1975). males are significantly different from those in the anuran species (8). The role played by 7. Physiol. C. D. Geisler, W. A. van Bergeijk, L. S. Frishfemales (P < .001, Kolmogorov-Smirnov twokopf, J. Morphol. 114, 43 (1964). the high-frequency fibers is not obvious. sample test). This test also revealed that the Q 8. R. R. Capranica, L. S. Frishkopf, E. Nevo, values for the low- and mid-frequency units Occasionally the male E. coqui emits a Science 182, 1272 (1973); R. R. Capranica, in were not significantly different in males and Handbook of Frog Neurobiology, R. Llinas and call whose function is unknown and females (P > .05). W. Precht, Eds. (Springer-Verlag, New York, in At least eight sympatric species of eleuthero15. which contains energy as high as 2700 press). dactylids contribute to the sonic background in best excitatory frequency of an auditory hertz (12). Perhaps the high-frequency fi- 9. The the El Yunque rain forest [A. Schwartz, Stud. nerve fiber is the tonal frequency to which the Faun. Curacao Other Caribb. Is. 91, 1 (1967)]. bers provide the basis for detection of fiber is most sensitive (that is, has the lowest 16. E. Batschelet, Animal Orientation and Navigathreshold). this call. The mean BEF of the low-fre- 10. A unit is inhibitable if its response to an extion (NASA SP-262, Government Printing Office, Washington, D.C., 1972), pp. 61-91. citatory tone within its tuning curve can be 17. The quency units is not significantly different + + response was a stronger indication of addition simultaneous the totally suppressed by female preference than a + response. The linear in males and females (P > .05); however, of a second tone of higher frequency and approdistance from the release point to the midpoint priate intensity (up to 30 db above the excitatory the differences between the mean BEF's of the + + response area is 1.75 m; the distance tone) outside the unit's excitatory tuning curve. to the midpoint of the + response area is 0.75 m. of both the mid- and high-frequency 11. In the bullfrog (Rana catesbeiana) the auditory The ratio of these two distances is 2.33. Assumnerve fibers which innervate the amphibian paunits are significantly different for the ing responsiveness to be a linear function of pilla are sensitive to low- and mid-frequency distance traveled by the female, we assigned a two sexes (P < .001) (13). tones and the fibers which innervate the basilar weight of 2.33 to each + + response and a weight to tones are sensitive papilla higher-frequency of I to each + response. The sharpness of tuning of an auditory [A. S. Feng, P. M. Narins, R. R. Capranica, J. thank G. E. Drewry, J. Col6n, and D. Comp. Physiol. 100, 221 (1975)]. A similar orga- 18. We nerve fiber is given by its Q value (the Narins for providing live E. coqui; P. Capranica nization is presumed to be the case for E. coqui. for aid with the field experiments; the Puerto BEF divided by the bandwidth of its tun- 12. G. E. Drewry, in Puerto Rico Nuclear Center Rico Nuclear Center for the use of its field Rain Forest Project Annual Report 1970 (Puerto ing curve at 10 db above threshold). The facilities; and the members of our laboratory for Rico Nuclear Center, San Juan, 1970), pp. 16comments on the manuscript. Supported by sharpness of tuning of the fibers sensitive 63. NIH research grant NS-09244 to R.R.C. and a 13. test used was the The Kolmogorov-Smirnov to low and middle frequencies was not Sigma Xi grant-in-aid of research to P.M.N. two-sample test [S. Siegel, Non-Parametric Stasignificantly different in males and fetistics (McGraw-Hill, New York, 1956)]. A 21 November 1975; revised 29 December 1975 males. However, the high-frequency units showed a distinct difference in tuning sharpness between the two sexes; for females, the average Q of the high-fre- Incentive Contrast in Honey Bees quency fibers was 0.91 whereas in males it was 1.34 (14). Abstract. Bees trained to come to the laboratory for a 20 percent sucrose solution The fibers sensitive to high fre- accept it readily, but bees trained with a 40 percent sucrose solution and tested with quencies in females tend to have lower the 20 percent solution show a pattern of interruptedfeeding that may last for several BEF's and broader tuning than those in minutes. Bees trained with 20 percent and tested with 40 percent sucrose are males. This indicates that the basilar pa- undisturbed. When the animals are offered two samples of the 20 percent solution pilla in the inner ear is tuned differently simultaneously, they drink to repletion from whichever they first taste on each visit, in males and females. Such sexual selec- but if both a 20 percent and a 40 percent drop are offered the 20 percent solution is tivity in the response properties of the rejected after a single experience ofthe 40 percent solution. Although these results are anuran's auditory organs is yet another analogous in many respects to incentive contrast effectsfound in mammals, they can adaptation for peripheral stimulus filter- be understood in sensory terms and do not require the assumption of learning about ing. Partitioning the call so that it serves reward. (at least) two functions, and differentially adjusting the peripheral sensitivIn diverse mammalian species, experi- which they have previously found both ities to enhance reception of the parame- ence with a preferred food may reduce grain and bread (3). These results have ters of the call of interest to each sex, the acceptability of a less favored food. been cited in support of the view that provides a novel mechanism whereby a Monkeys will eat lettuce from a cup they rewards do not serve simply to strengthmale can simultaneously communicate have earlier seen baited with lettuce but en response tendencies as reinforcement with both sexes in a complex sound envi- not from one earlier seen baited with theorists have proposed (4), but that reronment (15). banana (1). .Rats trained with bran mash wards may be learned about, and anticiPETER M. NARINS in a maze or a problem box will not pation of their specific properties may ROBERT R. CAPRANICA accept sunflower seeds substituted for play an important role in instrumental Section of Neurobiology and Behavior the mash, although sunflower seeds are behavior (5). Incentive contrast does and School of Electrical Engineering, readily accepted by rats trained with not, however, necessarily imply leaming Cornell University, them from the outset (2). Goats will nor- about reward. In honey bees (Apis melIthaca, New York 14853 mally eat grain, but not in a situation in lifera), I have found contrast effects, di-

380

SCIENCE, VOL. 192

rectly analogous to the mammalian effects, that can be explained in simple

tested with the higher were entirely undisturbed; the marked difference in besensory terms. havior (P