physidogy & Behavior. Vol. 45. pp. 1087—1092. e Maxwell Pergarmn Macmillan pic. 1989. hinted in the U.S.A.mu.9384Æ9 $3.00 + .00
Behavioral Determination of Critical Flicker Fusion in Dogs
D. CAROLINE COILE, I CELIA H. POLLITZ AND JAMES C. SMITH
Psychobiology/Neuroscience Program, Florida State University, Tallahassee, FL 32306-1051
Received 19 December 1988
COYLE. D. C.. C. H. POLLITZ AND J. C. SMITH. Behavioral determination ofcriticalflickerfusion in dogs. PHYSIOL BEHAV 45(6) 1087—1092, 1989. —Steady-state critical flicker-fusion frequencies (CFFs) were determined for four beagle dogs using the psychophysical technique of conditioned suppression. ne CFFs obtained demonstrated that the dog can discriminate flicker at much faster rates than has been suggested by ERG data. In addition, dog rods may suppon the discrimination of flicker at much higher rates than can human rods. An indication of a psychophysical rod-cone break occurred at a luminance level intermediate to those previously reported in ERG CFF studies. This level is similar to that in the cat, but much higher than that in man. The high CFFs provide the first psychophysical evidence of a well-developed functioning cone system in the dog.
Dog (Canis familiaris)CanineCFF FlickerAnimal psychophysics Rod-cone break Photopic
Scotopic
ALTHOUGH anatomical and electrophysiological studies have indicated that the dog has a fairly wen-developed cone system (27,37), no behavioral studies have demonstrated the extent of the doss photopic vision. The time involved in training animals to make visual discriminations. coupled with the assumption that the results will be similar to electrophysiological measurements. makes psychophysical measurements extremely scarce for most species. In the dog, a particularly important animal model for many ocular disorders (19). there is surprisingly little normative psychophysical visual data of any type available. One of the most commonly employed electrophysiological assessments of visual function in dogs. the critical flicker-fusion frequency (CFF) based on electroretinograms (ERGs). has only one behavioral data point for comparison (32). The present study was undertaken to provide normative psychophysical CFF thresholds for the dog, with special There' is very little behavioral corroboration for CFFs in dogs, however. The single published psychophysical study (32) used only one stimulus light intensity. The CFF threshold obtained was much higher. indicating more sensitivity to flicker. than CFFs reponed from ERG studies. In CFF studies with humans, it is not uncommon to obtain higher subjective CFF thresholds than can be observed through monitoring ERG fluctuation (10, 15, 21). The situation is not very clear. however, when comparing the results of animal psychophysical and electrophysiological studies because the reliability and sensitivity of various psychophysical methods may influence reported thresholds.
generally increases with increasing light intensity, in animals with a functionally duplex retina the CFF increases at rates different for scotopic levels of illumination than for photopic levels. CFF thresholds plotted as a function of stimulus intensity should / olfactory detection studies with dogs (28) and in CFF studies with monkeys (39), cats (30). rats (43), and pigeons (22) suggested conditioned suppression as the method of choice for CFF determination in the dog.
has been thoroughly described in humans (40) and investigated in many other species (16.23). Although CFF ). The reliability of this technique in
produce a nonmonotonic curve (with a rod-cone break) for animals with duplex retinas when a broad range of stimulus intensities are employed. CFF curves obtained from electroretinograms (ERGs) give a clear indication of a rod.cone break in normal dogs (3, 4.
37. 38) although an early investigation failed to elicit a cone portion (36).
*Requests for reprints should be addressed to D. Caroline Coile. Department of Psychology.
1087
A recurring problem in animal psychophysics is the tendency of the animal to make responses even in the absence of the S + (stimulus to be reinforced). Instead. it appears to be more reliable to train an animal to discontinue responding in the presence of the
METHOD
Subjects
Four beagles (one male and three females). ranging in age from three to six years, were used in the study. Examination by direct ophthalmoscopy revealed no evidence of ocular abnormalities in any dog. Because we were interested in the dogs' natural ability to detect flicker, pupils were never artificially dilated for test sessions.
Dogs were housed in indoor-outdoor runs (6.5 x 1.0 m). and so were subject to natural lighting and temperature conditions. Water was available at all times except in the experimental chamber. Beef baby food obtained by the dogs as reinforcers was supplemented by a daily meal of dog chow given 30 minutes after the end of each training session.
Apparatus
Behavioral The experimental chamber was 1 17 cm in height, 105 cm in length. and 73 cm in width, with a floor made up of metal bars which served as a grid for electric shock. Dogs were free to move about the chamber. The entire chamber was situated inside a darkened room, which was separate from the room housing recording and control equipment.
A teflon nose-key (7 cm in depth by 5 cm in width) was recessed at one end of the box, and was attached to a microswitch which closed when the nose-key was pushed upward about 0.5 cm, A "paste feeder" (9) dispensed baby food reinforcements 21 cm below the nose key. Food was pushed through a 1 . 1 cm aperture into the dog's chamber, and the size of each reinforcement (usually about two to three grams) could be controlled by the experimenter.
Current for the chamber•s grid floor was generated by a Lehigh Valley Constant Current Shocker (model 122-04).
Sttmttttmlighmtourca. On each side of the dog's head, about 5 cm from the nose-key and 17.5 cm from the walls of the box, were vertically mounted 22.5 cm fluorescent gibes. The tubes were situated at approximately a 45 degree angle with respect to the sides of the box. in order to present the light as nearly as possible along the dog•s optic axes. When a dog was pressing the response key with its nose, its eyes were approximately 14 cm from the filters in front of each tube. Figure I diagrams the relative positions of the stimulus light sources, nose-key, paste-feeder. and responding dog.
The fluorescent tubes were ED-3 lamps with a high speed ( < 100 microseconds) phosphor, and were energized by a high compliance constant current source to stabilize light output. The spectral output of these tubes was elevated in the blue end of the spectrum, which has been shown to facilitate the appearance of a rod-cone break in CFF intensity curves (20). Directly in front of each tube was a 6.25 cm wide by 25 cm high area of Kodak Wratten eeiatin filters mounted between two pieces of ground glass. with edges bound by opaque black tape. Neutral density filters of 0.3 D. 1 .0 D, and 1 .3 D were used in combinations to obtain stimuli intensities of 1209. 557, 188. 57. and 5.8 cd/sq-m [measured using a Tectronix J 16 photometer with the J6503 (ftL) probe placed as nearly as possible in the position of the dog •s eyes when the dog was pressing the nose-key]. These values were chosen to bracket the intensities at which most ERG CFF rod-cone breaks have been reported for the dog (2—4. 37).
Fiuotescent tube (behind ground gloss)
FIG. l. Experimental chamber showing stimulus lights. nose-key, paste feeder. and dog. Fluorescent tubes (shown in dashed lines) are not actually visible.
at •13&Hz:• and had i the:
Procedure
The probability of reinforcement was gradually decreased to a 0.02 variable ratio schedule. Throughout operant training the stimulus light was set at an intensity of 1209 cd/sq-m and a flicker rate of 130 Hz, and was assumed to appear as fused light to the dog.
CFF [N DOGS
2.n Strip chartrecord i0f one subject at the maximunuilluminauon of 209,5 cd/sq-m, Trials are presenteå in the-order imwhich they tw•curred. With intertfial intervol&, (averaging*45 seconds) omitted from the record.
( 3 ) , *Ar resulting value op. 1'.0 indicates- completesuppression. and indicates.no suppression. An SR of (Y:33 :aæwhich indicates that half •asqmany •responser•inæ the e• period as in the P period*30 was. valtr.
Amplitude of the shock was determined individually for each dog in an effort to keep the level as low as possible without losing effectiveness. It was considered to be sufficient when it elicited a consistent startle response from the dog. Effective levels ranged from 0.3 mA to 2.5 mA. These levels were not considered to be painful by humans touching the grid floor with the palms of their hands.
pp'ox imately• I fl ickeetrials*l O»sharmtrials;, an&i Obasehne trials-were rune dailyeTria19*weve-initiaed
Initial training and data collection were with 1209 cd/sq-m stimuli; later data were collected with 57, 5.8, 1 18, and 557 cd/sq-m stimuli. in that order. These levels were selected to provide an adequate range in which to plot a rod cone break.
RESULTS
All subjects attained mean SRS greater than 0.80 by the third session of suppression training. with flicker frequencies of 10. 2().
1089
.00
0557 0
1209.5 cd/sq m
o
0.80
o
c 0.50 o
0.40
SR-O.JJ
0.20
0.00
-0.20
o206080too
Flicker Frequency (Hz)
MeaibSR9* aFzrfunctK"t OF flicker- frequemyaedifteren•-sai.mulff, gntensi'ies Dashed tines connectthe SRS, soiid imes are lea* squares4tiL,oi, SRS lying. be'*eeodashed" line at denotes the threshotdt eriteriöik
and 30 Hz. Shock ievel.s were stabilized at ().3 for two subjects (Liz and Tater). and 2.5 BIA for the other two subjects (Cleo and Tara). This wide range was due to the different reaction of the dogs to the shock. so that sorne dogs required a higher level of shock to elicit a startle reaction or suppress responding.
As demonstrated in conditioned suppression studies with cats (30) suppression behavior was not all-or-none. but instead exhibited a graded response across a range of frequencies around threshold. For example. a strip chart recording of a typical
Average baseline and sham trials for all of the dogs indicated a slightly higher rate of suppression for sham trials (approximately 0, 12) than for baseline trials (approximately 0.05). It was not uncommon for dogs to pause for a second after the black-out period and appear to attend to the stimulus lights before resuming responding during a sham trial. The fact that both baseline and sham means are above 0 is a result of the practice of aborting ail trials in which no responses occurred in the last 5 seconds of the P period. thus reducing the probability of negative SRS.
Mean flicker trial SRS as a function of flicker frequency for the various intensities are presented in Fig. 3 for one subject. Similar data for each subject were used to determine CFF at each stimulus intensity by fitting least-squares regression lines to SR value.s lying between 0.80 and 0.20 (see Fig. 3). Above 0.80 and below 0.20 the SRS tended to stabilize. indicating either final suppression levels or initial baseline (no suppression) levels of responding. CFF as a function of intensity is plotted for each subject in Fig. 4; three of the subjects had very similar thresholds, while one (Tara) tended to have lower thresholds throughout all intensities.
too
A Liz
O Tara
Toter
80
c
Q)
60
20
110 3
Luminance (cd/ sq m
CFF as a function of stimulus intensrty for each subject'.
DISCUSSION
Compär!sons : With ERG Siudies and Other Species*
Comparison of psychophysical with published electrophysiological thresholds are of special interest in order to evaluate the extent to which the ERG, which is the most popular means of evaluating retinal function in veterinary ophthalmological research, agrees with the dog's actual ability to detect flicker. -In— ERG studies of dog-CFÄ3, 4, 37, 38),
(Fig. 5). This is in agreement with the previously reported psychophysical CFF estimate at a single intensity (31) (Fig. 5).
Ourr resu ts- suggesc—a.-rod-
CP&S)
psychophysical studies [LdS: (32). CP&S: present study]. Curves from studies other than the present study may not be shown in their entirety.
FIG. 6. Psychophysical CFF curves for the dog (present study), cat [LP&S. (31), K: (25)), monkey (39), and rat (43). Curves from studies other than the present may not be shown in their entirety.
which is intermediate to the levels obtained in ERG studies, and close to the psychophysically determined level for the cat (31) (Fig. 6). This is much higher than the luminance level at which the human rod-cone break occurs [around O. I cd/sq-m (12)]. As in the
(IQ)}. Such high rates of flicker detection might raise the concern that the tested stimulus luminances were not dim enough to selectively stimulate the rods. so that the curves would represent only cone responses, despite the fact that the lower intensities tested were well below that at which the rod-cone break has been shown to occur in most dog ERG CFF studies. Loop et al. (31) tested their cats at luminance levels as low as —4.5 log cd/sq-m and still obtained CFFs of 20 Hz. They used several techniques to demonstrate that the cat's rods were responding to high rates Of flicker, including eliminating the rod branch of the curve by exposing the cats to rod-saturating adapting lights between trials.
æGFEincensitrcurve•whicWdnpoff-aehighepintensities (43). The comparison with the rat CFF is of special interest because the stimulus presentation apparatus (with a different range of light intensities) and the psychophysical methodology were virtually identical to those in our dog study.
According to Dodt and Wirth ( 1 6), the intensity of light needed for the photopic branch of the CFF curve is related to a species' rod-cone ratio. This is clearly not the case in Fig. 6, as the dog and cat both have about 5% cones (35.42), which is less than in the monkey (14% perifoveally (I)) but greater than in the rat [about 2% (29)]. Comparisons of ERG CFFs as a function of retinal illuminance in several species have similarly failed to find a relationship with rod-cone ratio in mammals (23).
The present study is the first to employ conditioned suppression
CFF IN DOGSSR=O.JJ
SR-0
SR-0.56
50
60 c
40
20
Luminance cd/ sc m
FIG. 7. CFF curves in the present study calculated using threshold critena of SR -0.33. 0.50.and 0.66.
to evaluate visual function in the does Although* our dogs maintaineå high-suppressior•ratios for the lower flicker frequencies-. u we founde from subthreshoid. and suprathreshokl i ry some species I (2 ) our dogexhi response- over the situatiomreported In comparison to ERG studies, the slightly more variable responses that are typical of any animal psychophysical study may have tended to obscure a rod-cone break in our data. This is especially true since the psychophysical data had to be collected over many sessions (using the Method of Constant Stimuli) for each subject, while the ERG data can generally be collected within one session by adjusting the flicker frequency to threshold for each intensity. In addition, the time involved in testing at each stimulus level made it difficult to include as many different luminance levels as we would have liked.
A problem in comparing thresholds obtained from different techniques is that the criteria for determining thresholds are difficult to equate. ERG CFF threshold is generally taken as that frequency at which the b•wave no longer appears to exhibit a correspondence with the flickering stimulus. (h•psychophysiR11 'hedoe maker haif*
our data using more stringent criteria of 0.50 and 0.66 SRS resulted in lower CFF thresholds in all cases. but in no case as low as reported ERG thresholds. and with less indication of a rod-cone break (Fig. 7).
Procedural Considerations
When making comparisons with ERG results severai procedural ditTerences should be considered. The ERG studies used artificially dilated pupils, while we allowed the pupil to remain in a natural. responding state. The use of dilated pupils would theoretically result in a rod-cone break at a lower stimulus intensity, the opposite of what we generally found. The ERG CFF rod-cone break determined by Dodt and Enroth (14) in lightadapted cats with dilated pupils was about one log unit higher than the psychophysically deterrnined rod-cone break found by Loop et al. (31) also using cats with dilated pupils.
Larger stimuli have been shown to result in rod-cone breaks at lower luminance level> in humans (40) but there is no consistent effect in the dog CFF data. Our stimulus subtended a larger visual angle than did that of Aguirre and Rubin (4) but not as large as the Ganzfield illumination of Schmidt (38) and Schaeppi and Liverani (37). Binocular viewing (as used in both psychophysical studies of dog CFF) has been shown to result in slightly higher CFF thresholds than does monocular viewing (as used in the ERG studies) (24).
CFF tends to decrease with an increasing ratio of li oht duration to dark duration (44). Aguirte and Rubin (4) used a light-dark ratio of I for their flicker stimulus. while the present study used a : t light-dark ratio. Neither Schmidt (38) nor Schaeppi and Liverani (37) specified their light-dark ratio. although a ratio is most commonly employed. Another difference between the ERG and psychophysical conditions is that the dogs are initially anesthetized. and then [except for Schmidt (38)] paralyzed. for the ERG recording sessions. It is possible that the CFF could be affected by these drugs (40).
Conclusion
The ability of the dog to discriminate flicker at higher frequencies than has been demonstrated in ERG studies provides the interesting speculation that dogs may make greater use of their photopic visual system than has been previously believed. Given the utility of the dog in many tasks requiring good vision, as well as its role in ophthalmological research, it may prove worthwhile to more thoroughly investigate the functional abilities of this animal's visual system.
The CFF intensity curves obtained in this study should provide a useful basis for comparison of visual ability in dogs with a variety of visual disturbances. Comparisons of anatomical, electrophysiological, and behavioral data in other species with ocular disorders have shown that in some disorders the techniques yield very different results. For example, it has been shown that anatomical evidence of photoreceptor damage is not necessarily accompanied by similar deficits in behavioral visual tasks in rats (5—8), ERG measurements, too, may not correlate with behavioral measurements. since the ERG may remain normal as long as receptor and bipolar function are maintained, even after vision has been lost. For example, central nervous system diseases which may affect the visual pathway also may not affect the ERG. One such example in the dog is ceroid lipofuscinosis, which is similar in many respects to Batten's disease in man (33). Also. ERG data do not necessarily reflect diminished visual perception due to brief retinal anoxia (34) or glaucoma (l I .26) in humans. Yet the dog is an important animal model for both glaucoma (18) and ceroid lipofuscinosis (33). Research in these areas should not neglect psychophysical measurements simply because of the greater ease of electrophysiological measurements.