456 AM J OPTOM & PHYSIOL OPTICS

Vol. 65. No. 6

trast) cues were superimposed; they were grad-

ually eliminated after correct responding. Even-

tually all monocular cues were removed, requir-

ing the patient to respond only to stereoscopic

cues. It was found that the percentage of correct

responses to stereo targets was improved dra-

matically with reinforcement. Most children

aged 3 years or older were able to respond süc-

cessfully to a RDS presented in this manner.

Nonreinforced responses to stereograms were

poorer and more variable. Cooper and Feldman2

concluded that RDS presented in an operant

conditioning paradigm were particularly effec-

tive in evaluating binocular responses and de-

tecting a constant strabismus because RDS do

not contain any monocular cues and require

bifoveal alignment for perception (Table 1).

In a later study, Feldman and Cooper modified

their automated operant RDS technique by uti-

lizing errorless discrimination along with cue

fading techniques.3 This time both the RDS

containing disparity (5+) and a RDS lacking

disparity (S—) were presented simultaneously

(Figs. 1 and 2). The left to right location of the

Si+ was altered randomly. Patient responses

were made by breaking an infrared photocell

beam when the child tried to touch the stimulus.

Both the contrast of the monocular cue in the

stereoscopic RDS and the contrast of the incor-

rect stimulus (no stereo, S—) were reduced; that

is, the monocular cue faded out while the con-

trast of the S— was faded in. The technique

enabled valid and reliable responses in children

as young as 2½ years of age. Traditional testing

techniques that required an experienced doctor

or technician resulted in no responses or unre-

liable responses in many children under 4 or 5

years of age.

TABLE 1. Number of patients passing or failing the

operant RDS dlscñminatlon test according to visuai

diagnostic dassification.

• Normal refers to patients whose visual diagnosis

did not include a strabismus (constant or Intermittent).

amblyopia. or ocular pathology.

Similar computerized techniques have also

been used by others. Fox et al.4 used computer-

generated dynamic RDS with a preferential

viewing procedure in order to investigate bin-

ocular responses in infants. Their subjects

viewed the dynamic RDS while wearing red-

green anglyph glasses. A stereoscopic vertical

bar was moved from the center position to the

left or right. A trained observer viewed the po-

sition of the infant’s eyes. Appropriate fixation,

i.e., movement of the eyes corresponding to the

position of the vertical bar, signified stereo-

scopic appreciation. Stereopsis was demon-

strated in infants as young as 6-months-old.

The experiments by Cooper and Feldman1-3

and Fox et al.4 demonstrated that appropriate

stimulus presentation associated with effective

reinforcement could be used to investigate bin.

ocular vision in patients who lacked sophisti-

cated communication skills. Both research

groups used computerization to present and ma-

nipulate stimuli, and to present reinforcement

when necessary. Computerization was required

to make rapid, almost instantaneous changes in

stimulus parameters and to provide immediate

feedback of reinforcement. Manual techniques

would have been too slow and arduous.

In another experiment, Cooper and Feldman5

used their operant conditioning techniques with

automated presentation of RDS to determine if

vergence training resulted in an increase in ver-

gence ranges. They used an A-B reversal design

to control for placebo effects. The experimental

group (A) received vergence training; the control

group did not (B). During vergence training,

correct responses resulted in positive reinforce-

ment and a concurrent increase in vergence

demand, whereas incorrect responses resulted in

a reduction of vergence demands. The control

group received the identical stimuli and rein-

forcement; however, neither correct nor incor-

rect responses resulted in any alteration of ver-

gence demand. Their results demonstrated that

automated convergence training yields a rapid

increase in maximum convergence range,

whereas placebo training does not. Furthermore,

patients who improved their vergence ranges

using this system transferred their ability to

other vergence tasks involving vectograms and

prisms. Cooper and Feldman also demonstrated

that patients who did not respond to traditional

orthoptic therapy were treated successfully with

automated vergence training (Fig. 3).

Recent research by Dawn et aL supported

Cooper and Feldman’s previous work that com-

puterized convergence training improves posi-

tive fusional vergences. Dawn et al. demon-

strated transfer of improved vergence abilities

on prism bar and amblyoscope testing devices.

In a later clinical study Cooper et al.7 designed