Lateralization performance of squirrel monkey (Samiri sciureus) to binaural click signals.

AN IMPORTANT FUNCTION of the auditory system is to determine the locus of a sound source. Two of the prim;iry <tcoustic cues for localization are the interaural time difference (/11) and the interaural intensity difference (/11). In man, the presentation through paired earphones of dichotic stimuli having interaural time separations of less than l or 2 msec gives rise to a uni tary aroustic image localized someplace "within the head." Under the.,c experi mental conditions, the a~!ics~ment o( the i ntracranial position of the acoustic image is a process ohen

AN IMPORTANT FUNCTION of the auditory system is to determine the locus of a sound source.Two of the prim;iry <tcoustic cues for localization are the interaural time difference (/11) and the interaural intensity difference (/11).In man, the presentation through paired earphones of dichotic stimuli having interaural time separations of less than l or 2 msec gives rise to a unitary aroustic image localized someplace "within the head."Under the.,c experimental conditions, the a~!ics~ment o( the i ntracranial position of the acoustic image is a process ohen referred to as "lateralization" (3).In lateralization experiments, in contrast to localization, it is possible to vary the two interaural cues independently.However, the neural processes invoked both in the localization of an externa l sound source and in the lateralization of an internal image are thought to be similar.Lateralization has been well studied psychophysically in humans (4, 5, 7-10, 13, I 8, 19, 21, 23, 27, 30) ancl neurophysiologically in cats (1 1, 20,22,24), and attempts to correlate the two types of data have been made (11,22).To avoid such generic differences there is a need for the collection of psychophysical data from animals amenable to neuropliysiological study.In the present paper we describe results of psychophysical lateralization experiments in squirrel monkeys using dichotic clicks that can he compared with a companion paper (25) in which si11glc-u11it response to the same click signals was obtained in another group of monkeys.
500-800 g.Their hearing, as jmlgecl from their ability Lo orietll toward low-level sound stimuli in Lhcir cn,•ironmenL, was appare11Lly normal.Each was trained in Lhe click-lateralizario11 task as follows (details of Lhe Lraining procedure can lie ol>tainccl from ref 6) : 1) The monkeys were habituated to siuing and eating in a restraining chair inside an acoustically souml-treated room and were never fed ehewhere.
2) They were trainee! to press either the right or lefL of two levers (Fig. I ) in order to receive a small (0. 19 ~) ha11a11a-ll;n•orcd food pellet.Each monkey co11sumecl approximately 100 of these nutritionally balancecl pellcLs per day, an amount which adequately supplied his caloric needs.
J) Once Jcyer-t rained .each animal, under anesthesia, recciYed a head mount attached to Lhe fromal hone of the skull with de ntal acrylic.,\ central ~c rew in Lhis device, when bolled to a frame-s upported cros~har on the restraining chair (Fig- .I), resLricted movements and allowed reproducible placemellt of earphones.
•/) Each animal was then hahilllated to pre~sing the le,•crs and eating while his head was bolted to the crossbar.
5) Each monkey was Lhen taught to press only one lever during 1he monaural preselllation of a train of clicks, and to achieve 85% correct responses in three consccuLive sessions.Each session consisted of at least JOO trials.The lever to he pressed was on Lhe same side as the earphone through which the click train was presented .The right-left sequence of stimulus presentation was randomly det ermined.The correct-lever response terminated the train of clicks and provided a food pellet.J\11 incorrect response lllrnecl o!I the lights in the sound room for 10 sec.
6) Trains of binaural clicks were then pre-sc11Led with varying-intcrau1•al Lime and in-tensiLy clifferences.The monkeys' task was to press the le,•er on the same side to which the earlier or more intense click of the binaural pair was presented.As in the mo11<1l1ral experiment, a correct response provided a food pellet and an incorrect response turned off the sound- room lights for lO sec.In presenting these binaural stimuli, the ear that received the initial or the more inte nse click was randomly determined.The monkeys' response terminated the click train an<l the trial.In a ny o ne session, either a time or an intensity difference was imposed , hut never both.Programming equipment (Iconix) was used to present the stimuli, record the re~ponses, ancl provide the appropriate reinforcements.

Sti11111li-confrol of j>arnmelers and methrul of presentation
The click stimuli were presented at a rate of 32/ sec ancl were produced l>y applying 250-~tsec pulses to 0.5-inch Bruel and K jacr condenser microphones (type 1134).The pulses (sec Fig. 2C) were monitored on a n oscilloscope to allow calibration of the pubc duration a nd inte nsity.As shown in Fig. 2, the acoustic outputs of the two microphones, as seen by a 0.25-inch condenser microphone set at a distance of 1.45 mm, are nearly idemical.At the maximum applied voltage used in this study (0 dB down), the click!. were approximately 60 dB above the experimenters' threshold.
When the monkey's head was rigidly fixed, the microphones were placed so that the housings were just touching the tragus of each ear (Fig. I).The microphone carriers were calibrated and their positions in the chair were recorded for each monkey to insure that the microphones would he placed in the same po~i tio11 frorn se;.sion to session.

Experiuwntnl procedure
Three lateralization experiments were performed on the monkeys at three intensity levels separated hy IO dB.In the first two expcrimems data were collccrcd relating the percentage of correct responses to the /::,.tand /::,.J values.In the third experiment, the end poim for 1at-erali1.ation was investigated.The eml poilll fo r A R, Mic.lateralization is the flt value at which the u11itary image appears to split a11d the click images arc resolved separately in each ear.
For a ll three experiments, a modified version of Wetherill a11cl Levitt's (29) hlock up-down method was used to cletermine the value of flt and fll from trial to trial in each session.I 11 this method, the flt a11d fll values are ranked to form levels of graded difficulty.With the exception of the encl-point study, it was assumed initially that the smaller the flt and fl! values, the more difficult it would be for the monkey to make a correct response.A block of trials was presented at the beginning of each session at some low level of difficulty (large b.t or fll value) and the monkey's performance over that block was recorded.If his performance exceeded, was exactly at, or was below the predesignated criLCrion, the succeeding block of trials was presented at a higher, the same, or a lower level of difficulty, respectively.The levels were varied in 20-µsec, 2-d B, and 500-scc steps for the flt, 6 1, and the end-point experiments, respectively.Blocks of three to six trials were used.The upper and lower hou11cls of criterion performance over a hlock of trials were clctermined by the formula given by Taylor and Creclman (26).ll should be emphasized that the purpose of using a modified hlock up-clown method was to provide a means for i.,rraclually approachi11g the cliffirnlt flt aml t.

M onnural task
All four monkeys learned the monaural task but each required vast ly different numbers of sessions and trials (100-200 trials/ session) to reach criterion perform• ance.Th e quickest mo nkey learned the task within 8 sessions (973 trials), while the slowest monkey required 35 sessions (4,8'!5 trials) .

At cx/J<:ri111cnls
Criterion performance o( 85% correct responses on the /it lateralization task was achie\'ed at values ranging from GO to 180 µsec and varied with the monkey and the intensity level.Figure 3 is a plot of the l:lt data from the monkey that performed in the most consistent manner.Th is monkey reached criterion performan ce at 60 µsec, with the highest level of cli ck stimu• lation used in this study.At a level 20 <lB less intense, criterion performance required the use of a larger l:lt value somewhere between 80-100 µsec.Thus, as the overall leve l of stimulation was increased, the monkey's p erformance, in general, increased for a given l:lt value.All monkeys who performed the /).t task at more than one intensity level demonstrated b etter performance as the intensity of the cli cks was increased over th e range o( intensities used in this study. Two-way analysis of variance on the last ~even sessions at each inte nsity level indicated that varying /).t and varying the overall intensity level significantly (a = 0.05) affected the correct-response probahili ties.Moreover, it was found that 20-µscc steps were marginal, and 40-µsec steps were usually s ufficie nt to significamly affect the response prubabililies with cofactors be ing the indi,•idual monkey and the overall intensity level.. lt should be noted that the .~i gnifi cant effects are restricted to /).t \'alues, which are not in the asymptotic region of the functions.As an overall comparison, Fig. 4 is presented to show lhe l.lt data for all four monkeys at one inte nsity level (0 dB d own for three monkeys and -I 0 dB down for the fourth monkey).

/).[ experim ents
Fig ure 5 is a plot of the per centage of correct lateralization as a function of Ill and overall intcnsily level over the last seven sessions for the same monkey whose /).l data were presented in Fig. 3. Her per-      formance was the best an<l most consistent of all the monkeys.Generally, all four monkeys reached criterion of 853 correct responses at a Lil value between 6 and 10 dB, and 2-dB steps were sufficiently large to affect the response probabilities.
Results of an analysis of variance test indica ted that there was no significant effect of click intensity on la teraliza tion performance based on Lil.Thus, there appears to be a differential effect of change~ in intensity level on lateraliza tion performance based on interaural time as opposed to interaural intensity disparities.Figure 6 is a combined plot of the L\l d ata for the four monkeys at one intensity level.
In comparing Fig. 6 with Fig. 4, it would appear that performance is much more consistent for all monkeys 111 the L\l than in the Lit experiments.
Trend analysis on the main effect of varying Lit or LH indicated that a linear component was significant, and that a l inear function accounted for more than 903 of the variation.

Percentage of 1•espo11ses lo right as a function of 8.t and Lil
The foregoing data have been presented in terms of percentage correct as a function of Lit and ~I.The data are presented m a slightly d ifferent manner in Fig.  in order to reveal some characteristics about responses and errors.The upper graph is a plot o( the percentage of responses to the right lever as a function of !!t, and in the lower graph as a function of !!I.It will be noted that there is a slight bias to the right.That is, at any of the three intensity levels, there is a tendency for the monkey to make more responses to the right lever at small !lt and !!I values, even when the click stimuli to the left ear led in time or were of greater intensity.Three of the four monkeys exhibited a right-lever bias, while the other monkey showed a left-lever bias.The response bias may be due to the fact that the monkeys actually perceived the image on the right side, or the task became difficult with the small (!lt and !!I) values and the monkeys developed a lever preference.Our results cannot distinguish between these two possibilities.Now, as the overall level of the clicks was decreased, there was an increase in responses to the right lever for left-leading stimuli.This increase in responses to the right lever with changes in intensity was not as striking in the AI data.Thus, there appears to be little change in the functions when intensity level is varied for the AI task but marked changes for the !lt task.

Lateralization end-point experiments
The initial block of trials was presented at a !lt value of 500 µsec and if criterion performance occurred, the 6.t value was increased by another 500 µsec.An example of the end-point data is shown in Fig. 8.For the three monkeys who performed this end-point task, performance fell below criterion at At values between 2 to 3 msec.
In Fig. 8 it will appear that decreasing the intensity by IO dB had no effect on performance when compared to the 0-dB level.However, reducing the intensity another I 0 dB (-20-dB level) caused a statistically significant decrease in correct responses resulting in the end point for lateralization being reached at a smaller llt value.Not all monkeys demonstrated a clear change in performance on the endpoint lateralization task with changes in intensity.

Generalization of localizing stimu./i
A basic assumption of this study is that the monkeys used the sidedness or laterality of the acoustic image as the primary cue for performance.1£ the assumption were correct, the monkeys should generalize to a localization task in which trains of clicks are presented free field through speakers (Altec 802D) to the right or left.A generalization test was performed on two monkeys.The speakers were placed at approximately 90° and 270° azimuths at a distance of about 50 cm.The generalization test consisted of either presenting the click stimuli through earphones at a 6.t value which the monkey had performed at least at criterion or through one of the two free-field speakers.For example, for monkey E the first IO trials were presented through the earphones at a 6.t value of 120 µsec.The ear which received the earlier click was randomly determined.In trials 11 through 35 the trains of clicks were presented through one of the freefield speakers selected in random sequence.For trials 36-49 the stimuli were again presented through the earphones at a At value of 120 µsec, and from trials 50-62 the clicks were switched back to the speakers.
Monkey E responded correctly to every trial in the sequence.It was further noted that the response to the first trial after each of the four transitions was not abnormal with respect to response time.It appeared that this monkey generalized to the localization task very well, supporting the notion that laterality of the acoustic stimuli was the primary cue used in the lateralization expenments.The second monkey (monkey J) did not generalize imme<liately but did learn the localization task within 50 trials.However, his data, as will be discussed later, must be interpreted with caution since he incurred an extensive neural lesion.

DISCUSSION
The present experiments show that squirrel monkeys can successfully perform a dichotic lateralization task using interaural time or intensity differences.Criterion performance (853 correct judgments) was achieved at /\t values of 60-180 µsec and 61 values of 6-10 dB.
Lateralization experiments in nonhumans are virtually nonexistent with the exception of a series of studies in cats by l\Iasterton and co-workers (11, J 6, 17) in which lateralization and its impairment following transection or ablation of various auditory structures were investigated.The cats were taught to generalize from a monaural click train to a binaural one with a large 6t of 500 µsec.Using a shock-avoidance technique and a flexible leather helmet to house hearing-aid earphones, they found that the threshold for lateralization (above chance performance) for normals averaged less than 50 µsec.No one has reported behavioral lateralization thresholds in monkeys.In human, Walsh (28) used the same stimulus-response paradigm as in the present experiment by having his subjects indicate wheLher the intracranial image was to the right or to the left.He reported that normal humans could seldom achieve 75% correct at /).£ values smaller than 70 µsec.However, Deatherage (4) states that humans can respond to binaural time differences of I 0 µsec.Observation from pilol work of the present authors indicates that humans frequently achieve 753 correct at 20 µsec an<l above chance at 10 µsec, but require ti1ne separations of 30-50 µsec to achieve 853 correct.It is seen in l'ig.3 that for one intensity, criterion was achieved at less than 60 µsec.Thus, the ability of squirrel monkeys Lo lateralize an intracranial image is close to the performance of humans on the same task.The differences that do exist may reflect the arduous experimental conditions to which the monkeys were exposed.the differences in one or more of the stimulus parameters (i.e., intensity level), and/ or the questionable reinforcing value of the food reward employed.
The notion, derived from both human psychophysical (19,30) and single-unit studies in animals (2), that intera ural time and interaural intensity parameters are processed differently by the nervous system receives support from three results of the present experiments.First, the more consistent performance of monkeys pre-:;ented with interaural intensity differences (see Figs. 4 and 6) may reflect the fact that i meraural time differences are minimized in an animal with such a small head size, and interaural intensity differences arc maximized for animals with such highfrequency sensitivity (15).Second, the fact that overall click level importantly affected lateralization judgments based on interaural time differences only (Figs. 3 and 5) is to be correlated with the single unit results described in the companion paper (see Fig. IO in ref 25) in which click intensity affected 6t sensitivity to a much greater extent than 61 sensitivity.And third, monkey ], whose left brain was atrophic at autopsy, performed the ~I task normally and the 6t task poorly.The lesion was ina<lvertantly produced when the head mount was implanted, and was primarily subcortical, particularly affecting the basal ganglia and internal capsule.The etiology of the lesion has not yet been established but appears compatible with a circulatory disturbance in the left hemisphere.The monkey showed the following symptoms after his lesion: J) paralysis of the right; 2) inability to localize tactile stimuli on the right; 3) inattention to the right spatial field, and 4) localization of external sounds to the left, that is, turning to the left when attempting to orient to any intense environmental sound.Heilman et al. (12) <lescribed similar symptoms in monkeys following ablation of the Lemporal-parietal cortex of one side.All of these symptoms disappeared within 2 months except for a residual slight paresis on the right that affected the use of the right hand.He adapted well to the training procedure and used his Jcf t hand to press the right aml left levers.He performed poorly on the 11t laLeralizaLion task and was unable to achieve criterion performance at the two lower intensity levels when 6.t values of less than 200 µsec were employed.His performance 011 Lhe I task was slighLly enhanced by intensity-level changes, a result not found for the other monkeys.More important, he was able to perform the 111 task as well as any monkey at the three inLensity levels.He <lid not generalize immediately in localizing external stimuli.It is possible, of course, that his poorer performance on Lhe 11t Lask was not related to the lesion but due merely to variation in the performance ability of squirrel monkeys.However, if the lesion did reduce his ability to use time as opposed to intensity cues, this would suggest that different mechanisms are utilized in the analyses of these two parameters.

Lateralization end-point cxjJerirnents
Observation from human studies (ref and preliminary studies) indicates that as 11t is increased Lo I.5-3.5 msec, with a consequent !>plitting of the fused acoustic image, the stimulus in the lagging ear is perceived to be less loud than to that in the leading ear.The time separation between the two ears must exceed by several milliseconds the 6.t at which the lag click or stimulus is first detected in order for it to appear as lou<l as the leading stimulus.
If the monkeys also perceived a loudness imbalance, they may still be able to perform well on the end-point task by utilizing the loudness differences.End point, in Lhis case, would reflect the time separation at which the lagging click is sufficiently loud to affect the loudness judgment, and not the separation at which the image begins to split.The decreasing function in Fig. 8 would repre~ent the effective loudness growth of the lagging click.If, on the other hand, performance begins to deteriorate as soon as the Jag click is perceptible, then it appears from the data that large 11t's are handled by both squirrel monkey and man in the same manner.
There are many variables to contend with when interpreting behavioral data from animals.The effects of the stressful experimental test situation, of changes in motivation, and of changes in temperament on test results must be large.
An experimenter never fully appreciates his subjects' preception or their strategy in responding to the task.These problems are reduced to some extent when verbal communication with subjects exists.The lack of such communication in the present experiment with !>quirrel monkeys has made us cautious in interpreting the data.

S U MMARY
Four squirrel monkeys were trained to make a right• or lcf t-levcr response depending on whether clicks were presented to a right or left earphone.They were then required to do the same for clicks presented binaurally when differences in the interaural time, intensity, and overall click level were introduced.Time and intensity differences and overall click level were varied in 20-µsec, 2-dB, and 10-dB steps, respectively.The percentage of correct lateral judgments was plotted as a function of interaural time and intensity differences for each of three click levels.Individuals achieved criterion performance (85 % correct lateral judgments) at time differences ranging from 60 to 180 µsec, and at intensity differences ranging from G to I 0 dB.The end point for lateralization was also examined for long time differences; performance fell below the criterion at interaural separations of between 2.0 and 3.5 msec, depending on the intensity level used.Changes in signal intensity affected performance on the 6.t but not the 111 task, suggesting that the neural mechanism subserving lateralization based on interaural time is different from that in which interaural intensity cues are used.
I levels and not to estimate a point 0 11 the psychometric function .The psy• chomctric fun ction is clcrivccl b y simply cal• culating the percentage of correct responses at each of the b.t and fll values.

2 ' 6 a 10 12 Differences
In di F IC.5.Percentage correct laLeral judgments as a fun ctio n of 6 1 at three intensity Je\'cls for monkeyE.

7
> leflF i e .7. PcrcenLage responses Lo the right lever as a funClion of l:lt (top) and M (bottom) for monkey E at three i11tcnsity lc"cls.
FIG.8.Lateralization end-point experiment.Percentage correct responses as a function of 6,t at three intensity levels for monkey E . .,,,'