Journal of Neuroscience Methods, 44 (1992) 145-155 © 1992 Elsevier Science Publishers B.V. All rights reserved 0165-0270/92/$05.00
145
NSM 01406
Assessing short-term recognition memory with forced-choice psychophysical methods A n t h o n y T. Cacace 1,2, Dennis J. McFarland 4, Joseph F. Emrich ~ and Jerome S. Hailer 2,3 Departments of l Surgery, 2 Neurology and 3 Pediatrics, Albany Medical College, Albany, N Y 12208 (USA) and 4 Wadsworth Center for Laboratories and Research, New York State Health Department, Albany, N Y 12208 (USA) (Received 6 January 1992) (Revised version received 20 July 1992) (Accepted 22 July 1992)
Key words: Short-term memory; Adaptive psychophysical methods; Memory span; Delayed matching-tosample The rationale and methodology for using computer-controlled forced-choice psychophysical methods to assess short-term recognition memory in human subjects are presented. Here, we use non-verbal computer-synthesized auditory and visual stimuli with an adaptive psychophysical procedure. Sequence-length thresholds (SLTs, span lengths) for randomly generated binary auditory and visual-sequential patterns and simultaneous visual-spatial patterns are determined to assess short-term memory capacity. The SLTs can also be used to equate for initial retention level for delayed matching-to-sample (DMS) or delayed matching-to-non-sample (DMNS) tasks which assess memory decay. The D M S / D M N S tasks have also been modified for use with the forced-choice paradigm. In contrast to many verbal paradigms requiring immediate ordered recall, non-verbal stimuli in a forced-choice paradigm provide a more direct measure of sensory memory because long-term memory, complex encoding/decoding processes, and motor-sequencing factors are minimized or avoided. Furthermore, the forced-choice recognition memory tasks are applicable over a broad age range, are less sensitive to socio-economic factors and educational level, and avoid complex instructions. Taken together, these factors enhance the applicability of these tasks in children and adults with CNS lesions, particularly where cognitive status may be compromised.
Introduction
Short-term memory assessment is an important component in neuropsychological or neurological test batteries which evaluate cognitive function (e.g., Demster, 1981; Franzen, 1989) and has been used extensively to study memory deficits after CNS lesions in experimental animals (e.g., Squire and Zola-Morgan, 1985; Columbo et al., 1990; Squire and Zola-Morgan, 1991). Their success in animal studies further enhances their
Correspondence: Anthony T. Cacace, Ph.D., Department of Surgery, Division of Otolaryngology, Albany Medical College, 43 New Scotland Ave., Albany, NY 12208, USA. Tel.: (518) 445-3713; FAX: (518) 445-5692.
value, which ranges from providing improved models of memory systems to the development of innovative clinical applications. Contemporary theories of cognition (e.g., Cowan, 1984, 1988) also use concepts of sensory storage mechanisms in short-term memory. The availability of low-cost microcomputers has made possible the cost-effective automation of procedures which assess sensory and cognitive functions. Combined with contemporary psychophysical methods, powerful tools are available to the behavioral neuroscientist to further advance memory research, to improve theory, and to provide technology transfer to the clinic. We report here methodology to assess aspects of short-term recognition memory in the auditory and visual modalities with particular emphasis
146 placed on obtaining sequence-length thresholds (SLTs, span lengths) using an adaptive psychophysical paradigm with non-verbal stimuli. Obtaining SLTs represents an important metric as it provides a direct index of short-term memory capacity. It also allows one to control for retention level in other short-term memory tasks such as the delayed matching-to-sample (DMS) or delayed matching-to-non-sample (DMNS) paradigms, which assess the decay of information in short-term memory. The importance of controlling for retention level when studying memory decay is well established (Underwood, 1966). Our methodology is very flexible, the tasks are easy to perform and they are potentially applicable to a wide range of subjects with many different levels of ability. We have found that these procedures can be successfully applied to young children (Cacace et al., 1990; Cacace and McFarland, in press) and in subjects with a variety of CNS lesions (McFarland et al., 1991). In addition, these procedures could easily be adapted for use with non-human primates. The system we will describe was developed around a commercially available product that is controlled by a microcomputer interface. At a conceptual level, we will break down the functions of the system into basic hardware components, stimulus synthesis, stimulus control, experimental design, psychophysical methods, and data tabulation.
Methods
Apparatus overL,iew The basic hardware architecture of the system consists of a 80286-based personal computer interfaced with the Modular Instruments M-100 system (Modular Instruments, Malvern, PA). This particular hardware has also been applied successfully to other scientific applications (Moore et al., 1989; Fuzessery et al., 1991). The Modular Instruments interface includes modules with 8 digital I / O lines (M-101A), 2 programmable stimulus generators (M-308), and a 2-channel attenuator-mixer (M-300). The PC configures the peripheral devices, controls real-time operations
with the Modular Instruments system and provides the interface for data collection. Also required is a mechanism to provide for recording subject responses and for providing feedback for a correct response. This can be accomplished either in the form of a mechanical voting (response) box, with light emitting switches, or with a computer monitor which incorporates a touchsensitive screen. The subject response interface (voting box or touch screen) is an addition to the Modular Instruments hardware. It should be noted that distinct advantages exist when using the touch screen over the mechanical voting box. The touch screen provides a convenient mechanism for presenting visual stimuli, provides the means for developing flexible reinforcement schemes which engage attention and maintain test performance level for subjects at different age levels, and also provides an easy means to study reproduction paradigms (i.e., motor-sequencing skill, including the measurement of reaction time). The software is written in Microsoft C. These tasks lend themselves to a design in which the interface with the experimenter, control of experimental trials, storage of data, etc., are all performed by common functions. Only functions which generate the specific stimulus elements (e.g., patterns of auditory tonal stimuli, visual colored geometric shapes, etc.), differ from one task to the next.
Acoustic stimulus synthesis The Modular Instruments system generates the auditory stimuli. To accomplish this, 1 cycle of a stimulus wave form is computed by the host computer and loaded into one or more of the programmable stimulus generators (M-308). Each M-308 module has 8 k of memory which can be expanded to 32 k. This generates a wave form with 12 bits of resolution. For the tasks discussed here, the synthesized stimulus wave forms are sinusoidal. Each stimulus generator is configured to recycle the wave form for a specified period of time. The stimulus generator modules also contain buffers which are loaded with digital rise/fall functions which can be programmed by the experimenter in response to specific needs. We
147
presently use the Blackman function (Harris, 1978) with 5-ms rise/fall times as a system default. A programmable clock is used to control the duration of the stimulus flat-time. Thus, stimuli of varying frequency can be digitally synthesized by varying the number of points in each sine wave which is then read out at a fixed rate (e.g., 50 kHz). The outputs of the 2 stimulus generator M-308 modules are input to the attenuator-mixer module (M-300) which regulates the sound pressure level of the wave forms and directs these signals to the appropriate output device (speaker or earphones). The digital a t t e n u a t o r / m i x e r is capable of modifying signals over a 100-dB dynamic range with a minimum step size of 0.375 dB. Etyrnotic E R 3 A insert earphones, are used to transduce stimuli independently between ears. The auditory stimuli consist of binary patterns which vary in terms of either frequency, intensity or duration. Each stimulus generator is loaded with the appropriate wave form and the pattern is then constructed by choosing 1 of 2 stimulus generators. The sequence is based on the bit pattern of a random number obtained from the random numbers generator of the Microsoft C library.
Vtsual stimulus synthesis For the short-term memory applications, we were interested in developing simple visual stimuli so as to allow comparison of performance in 2 separate sensory modalities with comparable tasks and stimuli. Thus, we generate visual stimuli of different colors, analogous to different frequency tones in the acoustic modality. The visual stimuli consist of binary patterns of 2 different colors (i.e., 2 visual wave forms differing in wave length). These are presented by means of a NEC multisync 2a analogue monitor which is interfaced to the computer by means of a V G A color graphics adapter. The graphics adaptor is set to 80 x 25 column color text mode, and the visual stimuli are presented as groups of solid characters (character no. 208 in the extended ASCII set) which vary in terms of the color attribute. Both visualsequential and visual-spatial stimuli are used. We typically use yellow and blue rectangles for both
visual-sequential and simultaneous visual-spatial pattern recognition memory tasks. The size of the rectangles are scalable and are under software control. For the visual-sequential task, we have found that 3.5 x 4.4 cm rectangles are easy to recognize and are suitable for our tasks. For the simultaneous visual-spatial recognition memory span task, the stimuli consist of a matrix of a variable number of 0.6 x 0.8 cm rectangles. The binary yellow and blue colors of the matrix are arranged as a random 'checkerboard' pattern. We chose these colors, since to our knowledge, a blue/yellow color blindness would be very improbable and thus our visual task should be applicable to the majority of subjects tested.
Experimental design Short-term recognition memory and psychophysical procedures. Our auditory and visual shortterm memory tasks share a common framework in that they are incorporated within a forcedchoice psychophysical paradigm and they use non-verbal auditory-sequential, visual-sequential, and simultaneous visual-spatial binary stimuli. Forced-choice paradigms are powerful since they are criterion free measures (Green and Swets, 1973). We use stimuli described above which are presented separately to the auditory and visual modalities, to determine SLTs, and to assess the decay of information over time with the DMS paradigm. Testing at least 2 different sensory modalities allows one to determine whether observed deficits are m o d a l i t y specific or supramodal in nature. Sequence-length thresholds (SLTs). Our current protocol for SLTs consists of a 3-interval 3 alternative forced-choice (3-IFC) adaptive psychophysical procedure. With the 3-IFC procedure, the subject is presented via earphones or observes via a computer monitor 3 separate binary pattern sequences, in each of 3 separate intervals. The duration and intensity of each stimulus element in the auditory/visual-sequential or visual-spatial pattern remains constant. On each trial, one randomly selected bit of a bit-pattern sequence, in 1 of 3 intervals, is inverted (i.e., changed to its alternative value). For purposes of simplicity and to avoid complex instructions, our
148
task was designed to follow an 'oddity' paradigm, with the odd pattern being randomly assigned to 1 of 3 intervals. The instructions to the subject are very simple (i.e., pick the pattern that's different). In this situation, the subject's task is to select the odd or different pattern in 1 of 3 intervals. To facilitate subject performance, the duration of each individual pattern sequence is cued by a light-emitting switch on the voting box or by a white rectangle on the touch screen. New patterns are generated on each trial, by a new draw from the random numbers generator. An example of the auditory-sequential task is presented in Fig. la and an example of the visualspatial task is presented in Fig. lb. All binary patterns are generated from the middle n bits of a 32-bit number obtained from the random numbers generator of the Microsoft C library. To acquaint the subject with the task as well as to determine if h e / s h e can discriminate the 2 stimuli, the length of the initial sequence begins at an easy level (i.e., at a sequence length of 2). Two
consecutive correct responses at this level increases sequence length, thus increasing task difficulty. One incorrect response decreases sequence length, in predetermined step sizes, and the task is made easier. This stepping rule, adapted from Levitt (1971), estimates an accuracy level of 70.7% correct in reference to the psychometric function. The SLT is determined from the subject's tracking performance and is the average of 'the last n-2' reversals. Feedback is provided for a correct choice. In our experience, we have found that an initial increment size of 3 elements and minimum step size of 1 element produces stable data. Our intention was to develop a task specifically designed to study short-term memory, that is to maximize stimulus uncertainty and minimize long-term learning of the patterns. By generating 'new' pattern sequences on each trial, practice or learning effects are minimized and short-term memory is emphasized. Thus, knowledge of previous patterns or use of learned verbal labels (i.e.,
Interval 1
Interval 2 Three Interval Forced Choice (Oddity Paradigm)
Inlervall
Jnterva[ 2
Time
knterval 3
Interval 3
m
m
Response Interval
Fig. 1. a: example of the 3-interval 3-alternative forced-choice recognition paradigm for use in the auditory and visual-sequential pattern recognition memory paradigm. The y axis represents the stimulus dimension (auditory or visual) and the x axis represents time of each pattern within an given interval. Three separate 7-element binary pattern sequences are presented as an example. In 1 of the intervals (i.e., interval 2), the pattern sequence differs from the other 2 by 1 randomly selected element (bit). The subject's task is to select the 'odd' or different pattern that occurred in 1 of 3 intervals, b: example of the 3-interval 3-alternative forced choice recognition memory paradigm using simultaneous visual-spatial pattern matrices. Each simultaneous pattern matrix, displayed as a random 'checkerboard' pattern (in this example black and white squares), occurs in each of 3 intervals. Three separate 15-element visual-spatial patterns are presented. As with the sequential pattern recognition task, 1 bit in the pattern is inverted to its alternative value producing a different pattern (i.e., interval 3). The subject's task is to select the odd interval by touching 1 of the solid black geometric shapes at the bottom of the figure to indicate h i s / h e r choice. In both examples provided above, the system uses a stepping procedure (Levitt, 1971 ) to adapt on the length of the sequence.
149 use of long-term memory) becomes irrelevant. Subsequently, for each trial, a new random number is selected and the process described above is repeated until a certain predetermined criterion is reached. Parameters such as characteristics of the auditory or visual stimuli (frequency/color, intensity, duration, etc.), interstimulus interval of elements within a pattern, intertrial interval between patterns, beginning difficulty of the task (i.e., number of binary elements in the pattern), stepping rules (number of binary elements to increase or decrease) or the number of reversals used to calculate thresholds, are all selectable by the experimenter and are stored as files. For ease of instruction and use over a broad age range and level of ability, the 3-IFC task follows an oddity paradigm.
Delayed matching-to-sample or matching-tonon-sample paradigm. The DMS task was also incorporated within the forced-choice paradigm. With this approach, a '3-interval 2-alternative' forced-choice methodology is employed with the same auditory-sequential, visual-sequential and visual-spatial patterns as described above. However, rather than adapting on sequence length, patterns of a constant length are presented an equal number of times/delay condition (i.e., randomized blocks design). In the 1st interval (i.e., the sample), the subject is instructed to remember a constant length binary bit pattern, consisting of either auditory-sequential, visual-sequential, or visual-spatial patterns, depending on the stimulus dimension being studied. Then, the subject is presented with 2 additional patterns in the 2nd and 3rd intervals. The subject's task is to select the bit pattern sequence or simultaneous spatial pattern matrix that occurred in the 2nd or 3rd interval, that is identical to the sample (i.e., matching-to-sample). Alternatively, if the DMNS paradigm is used, the subject would select the pattern that is different from the sample. In either case, use of DMS or DMNS paradigms is a software-selectable option. The patterns in the 2nd and 3 r d intervals are separated from the sample patterns by different silent delay intervals. Our silent delay intervals have ranged from 100 to 10,000 ms. The number of delay intervals, the
duration of the delay, the number of presentations/delay interval, and the length of the bit pattern are experimenter selectable. In all instances, they are presented randomly and performance is evaluated in terms of percent correct. In summary, application of forced-choice adaptive or randomized blocks techniques are means to study short-term memory using tasks that have proven to be useful both clinically and experimentally. These tasks were developed with the intent of improving the efficiency and accuracy of our test protocols. Since we are interested in studying humans, particularly children and adult subjects with various brain lesions, or other neuro-developmental problems, 'time' and 'accuracy' are primary concerns. In the following section, we will provide examples of results obtained from 2 subjects with similar lesions of the temporal lobe.
Results
The SLTs (span lengths) of 5 control subjects on the auditory-sequential, visual-sequential and simultaneous visual-spatial tasks are shown in Fig. 2a. Delayed matching-to-sample data, shown in Fig. 2b, are plotted in terms of percent correct with delay intervals ranging from 500 ms to 10,000 ms. The tasks were performed at the subjects' SLT. Span-length data of the subjects with documented CNS lesions, which are discussed below, are expressed as Z scores relative to this control data. Fig. 3a shows span length data from a 69-yearold man with a glioblastoma multiform of the right temporal lobe who underwent surgical resection. The resection margins includes the entire temporal lobe with sparing of the mesial amygdala and hippocampus with the posterior margin of the resection 5 cm from the tip of the temporal lobe along the 1st temporal gyrus and 6 cm along the 2nd and 3rd temporal gyri. A MRI image from this patient after surgery is shown in Fig. 3c. When the patient was tested postoperatively, a marked deficit on the visual span task was seen with near normal performance on the auditory span task. The DMS data, shown in Fig. 3b,
150 40 'A 35 r"
.+-
~
Visual spatial Visual sequential IF~[_q Auditory sequential (left ear) Auditory sequential (right ear)
30
(3) r- 25 O3 J 20 tO 15 [3_ (/I 10
Visual
Audifory
Sensory Modality
1O0
~6 L-
0 0 #t
90 80 70 60 50 40 50 20 10 0
O ~
" O ~
~
O------___ -O ,
O--O |--| V--X7 /',--A
Visual spatial Visual sequential Auditory sequential (left ear) Auditory sequential (right ear)
and 3rd temporal gyri, the amygdala, the anterior 1.5 cm of the hippocampus and the mesial temporal lobe to 6 cm from the temporal tip. A MRI image obtained after surgery is shown in Fig. 5c. The patient performed within the normal range on the visual span task and considerably below normal on the auditory span task. DMS results showed slightly abnormal auditory memory decay at the longest delay interval (Fig. 5b). The actual tracking performance of this subject is shown in Fig. 4b. As the examples illustrate, markedly different results can be obtained depending upon the modality of the stimuli used to determine SLTs. This is true even with lesions of the temporal lobe which appear to be grossly similar. Auditory verbal short-term memory deficits have previously been described in patients with parietal lobe damage (Warrington et al., 1971). Likewise, Hanley et al., (1991) have described a specific deficit in recalling non-verbal visual information. Such results provide compelling support for memory span testing in several sensory modalities.
Discussion 500
1000
5000
10000
Delay (ms) Fig. 2. A: control data (mean and SD) from 5 normal adult subjects showing SLTs (span lengths) f o r visual-spatial, visual-sequential and auditory-sequential pattern recognition memory tasks. B: control data showing performance for a DMS paradigm using the same visual-spatial, visual-sequential and auditory-sequential patterns. Four delay intervals ranging from 500 ms to 10,000 ms were used on the x axis. Ten presentations/delay interval were used. The number of elements in the pattern was set at the individuals SLT. Performance was evaluated in terms of percent correct, as shown on the y axis.
failed to show any significant memory decay over the period of 10 s. The actual tracking performance of this subject is shown in Fig. 4a. Fig. 5a shows span-length data from a 19-yearold female with a mesial right temporal and hippocampal low-grade astrocytoma who underwent surgical resection. The resection included the 2nd
In each case described above, the tracking data indicate that subjects converged on a threshold value during the course of the session. This illustrates the utility of the adaptive tracking procedure as applied to determining SETs. It should be noted that a considerable amount of analysis has been given to the development of adaptive tracking procedures, spanning well over three decades (e.g., Zwislocki, 1958; Levitt, 1971; Green, 1990). Hence, this is a powerful technique to employ, given that the performance metric can be thought of as a threshold. Quantitative psychophysical procedures can be readily applied to recognition memory paradigms, since recognition memory can be viewed as a specific case of discrimination. In this particular case, the stimuli are not all present at the time that a response is required. This view is assumed when signal detection theory is applied to the analysis of recognition memory experiments (e.g., Parks, 1966; Ringo, 1988). It should be noted, however, that
151
4C
3
A
I
100
Visual Spatial Auditory sequential (left ear) Auditory sequential (right ear)
B
_13
•
.I
90 BO
2
~,- 70 o K.
(-
Ef~ c J c o o.. V}
0
0
-1
60
50
(-) 40
~
-2
30
0--0 D--D ~7--v
20
-,5
Visual spatial Auditory sequential (left e a r ) . Auditory sequential (,right ear)
lO o
-4 Visual
Auditory
Sensory Modalify
,
-
500
.
.
.
.
.
,
.
.
1000
.
.
.
.
i
10000
Delay (ms)
Fig. 3. A: auditory-sequential and visual-spatial spans from a 69-year-old male who had surgery for glioblastoma multiform of the right temporal lobe. B: auditory-sequential and visual-spatial D M S performance at 4 delay intervals ranging from 500 ms to 10 s. C: postoperative M R I image showing the results of a right temporal lobectomy, see text for details.
152
AUDITORY SEQUENTIAL .... i
Left ear Right ear
o
.
VISUAL SPATIAL 30[ Binocular
.j\/.\,,
(to
~C~ C --I C O El. (,q
'~f\
Patlln|: 1
o
Patient: 1
/
~
i
20
30
40
i
'°[ ^
AA
/~,
~
/lo
Patient: 2
Patient:
2
o s
i0
15
20
25
30
35
40
0
10
so
60
Trials
Fig. 4. Patient l: tracking data from the 69-year-old male described in Fig. 3a for auditory-sequential and visual-spatial stimuli. The y axis represent span length and the x axis represents the number of trials to obtain SLT. Patient 2: tracking data from the 19-year-old female described in Fig. 5a for auditory-sequential and visual-spatial stimuli. The y axis represents span length and the x axis represents the number of trials to obtain SLT.
recognition memory experiments have not always employed the most sophisticated or rigorous of psychophysical procedures, and adopting a technique such as we describe here may provide a distinct improvement in this area of testing. With the adaptive tracking procedure, only 1 point on the psychometric function is obtained. This is in contrast to generatirig a complete psychometric function, which is more complete but also much more time consuming. While obtaining only 1 point on the psychometric function may be considered a limitation, the adaptive forcedchoice paradigm has other advantages. In particular, the threshold value obtained avoids floor and ceiling effects and it is a very efficient technique. We have also successfully applied the forced choice recognition paradigm to the study of other important short-term memory features such as serial position effects (Cacace and McFarland, 1992; McFarland and Cacace, in press). Furthermore, forced-choice paradigms are advantageous in that they minimizes motor demands from the subject. As previously mentioned, many traditional short-term memory tasks have used verbal
material and require immediate 'ordered' recall. Thus, when errors occur in the immediate ordered recall task, interpreting results can be problematic since one cannot easily distinguish between a recall deficit or one involving motorsequencing (reproduction) ability. This point serves to emphasize the fact that recall tasks also measure production deficiencies and that recognition tasks provide a less complicated index of short-term memory (e.g., Estes, 1982). This limitation is further emphasized when recall tasks are used to evaluate subjects with various CNS lesions. For example, motor-sequencing errors can occur with certain CNS lesions (e.g., lesions of the left frontal lobe), even when the memory component is controlled (Harrington, 1991). Given both advantages and disadvantages, we favor the forced-choice adaptive paradigm realizing that if these tasks are to be used clinically, they are preferable over lengthy tests which may actually produce data of lower quality, particularly if they exceed the attention span of children and adults with brain damage. This concern represents a distinct possibility and should not be overlooked or minimized. Ringo (1988) has examined a number of recognition memory studies which involved experimental lesions of the hippocampus in non-human primates. Re-analysis of the effects of hippocampal ablation in terms of d', a quantitative metric used in signal detection theory (e.g., Green and Swets, 1973), suggests that discrepancies between studies may have been due to differences in task difficulty. In a subsequent analysis of pertinent studies involving a larger variety of lesions (Ringo, 1991), when performance was also expressed in terms of d', many studies which appeared to demonstrate delay-dependent effects showed similar deficits at all delay intervals. This suggests that the effect of delay may simply be to make the task more difficult. The methodology described here insures that the subject is tested at a level approximating 70.7% correct, and methods are available that allow testing at any desired level of difficulty (Green, 1990). Thus, problems of sensitivity due to an inappropriate level of difficulty can be avoided. This method thus provides an efficient means of testing for delay-de-
153
5
15
100 B 90
- A E ; ~ Visual Spatial
It}
~V_I Auditory sequential (left ear) Auditory sequential (.right ear)
80
2
g)
1
c"
0
c fl)
-I
.4o I_
-2
70 60
0 O
50 40
N
3o 2o
c o -3 o_ 03 - 4
• • Visual spatial [] .D Auditory sequential (left e a r ) . ~ 7 - - ~ ' Auditory sequential (,right ear)
lO o Visual
Auditory
Sensory Modalify
500
1000
10000
Delay (ms)
Fig. 5. A: auditory-sequential and visual-spatial spans from a 19-year-old female who underwent surgery for a mesial low-grade astrocytoma of the right temporal lobe. B: auditory-sequential and visual-spatial DMS performance at 4 delay intervals ranging from 500 ms to 10 s. C: postoperative MRI image showing the results of a right temporal lobectomy, see text for details.
154
pendent effects that are not complicated by differences in level of difficulty. The methodology we describe could readily be adapted for use with non-human primate testing. For example, our visual-spatial task could be considered as analogous to visual-delayed matchingto-non-sample tasks that use 'junk' objects. In this case, the computer generates a video 'junk' object, the complexity of which is numerically specified. This methodology thus could potentially facilitate comparisons with human and subhuman primate data. Finally, we also wish to emphasize the importance of testing subjects in multiple modalities. This is particularly relevant in delineating such entities as a central auditory processing disorder (CAPD). This disorder implies difficulty in processing information presented specifically to the auditory modality, which is generally not attributable to peripheral hearing loss. To establish the existence of a CAPD, one must demonstrate that observed deficits are indeed modality specific. A recent paper reports success in segregating subjects with 'cerebral' lesions from normal subjects or subjects with cochlear hearing loss (Musiek et al., 1990) using a 3-element duration pattern 'reproduction' task which employed binary auditory stimuli of different durations. In contrast to the 'recognition' paradigm we employ, Musiek et al., (1990) asked subjects to assign verbal labels to the stimuli and 'reproduce' the order of the sequence verbally. However, due to the large variation of type and location of lesions that the study employed, the fact that only 1 modality was assessed, and given that a reproduction task was used, it is difficult to determine if these effects are specific to the auditory system or represent a more generalized deficit. The basic tenet of modality specificity requires that relevant comparisons of similar tasks be made in at least 2 major sensory modalities, Optimally, such multimodal tasks must be comparable, sensitive, and specific to the processing of information by central auditory or visual structures but insensitive to other factors, such as motor-sequencing abilities. If one cannot demonstrate modality specificity, the diagnosis of a CAPD remains uncertain. Consequently, alternative explanations such as a poly-
sensory or supramodal attention deficit disorder (e.g., Mesulam, 1981) must be explored. Furthermore, incorporating tasks within the forced-choice paradigm also helps to improve test specificity since it reduces motor demands and thus eliminates certain alternative explanations. The shortterm memory tasks described herein appear well-suited for this type of evaluation. In summary, this methodology allows for flexible, automated testing of auditory and visual short-term recognition memory. The nature of the stimuli allow for an adjustment of the difficulty level of the task in a quantifiable manner across a broad range. These procedures provide efficient techniques to investigate problems as complex as the effects of temporal lobe damage.
References Cacace, A.T. and McFarland, D.J. (1992) Modality effects in short-term auditory and visual pattern recognition memory: use of non-verbal stimuli. Assoc. Res. Otolaryngol.. 15: 69. Cacace, A.T. and McFarland,. D,J. (in press) Acoustic pattern recognition and short-term memory in normal adults and young children. Audiology. Cacace, A.T., McFarland, D.J and Wilder, D. (1990) Shortterm acoustic recognition memory in normal adults and young children. Paper presented at the Annual Meeting of the American Speech-Language-Hearing Association, Seattle, WA. Columbo, M., D'Amato, M.R., Rodman, H.R. and Gross, C.G. (1990) Auditory association cortex lesions impair short-term memory in monkeys. Science, 247: 336-338. Dempster, F.N. (1981) Memory span: sources of individual and developmental differences. Psychol. Bull., 89: 63-100. Estes, W.K. (1982) Learning, memory and intelligence. In: R.J. Sternberg (Ed.), Handbook of Human Intelligence, Cambridge University Press, New York, pp. 170-224. Franzen, M.D. (1989) Reliability and Validity in Neuropsychological Assessment. Plenum, New York. Fuzessery, Z.M., Gumtow, R.G. and Lane, R. (1991) A microcomputer-controlled system for use in auditory physiology. J. Neurosci. Methods, 36: 45-52. Green, D.M. (1990) Stimulus selection in adaptive psychophysical procedures. J. Acoust. Soc. Am., 6: 2662-2674. Green, D.M. and Swets, J.A. (1973) Signal Detection Theory and Psychophysics. Robert E. Krieger, Huntington, New York. Hanley, J.R., Young, A.W. and Pearson, N.A. (1991) Impairment of the visuo-spatial sketch pad. Quart. J. Exp. Psychol., 43: 101-125.
155 Harrington, D.L. and Haaland, K.Y. (1991) Hemispheric specialization for motor sequencing: abnormalities in levels of programming. Neuropsychologia, 29: 147-163. Harris, F.J. (1978) On the use of windows for harmonic analysis with the discrete Fourier transform. IEEE Trans. Acoust. Speech Signal Proc., 66: 51-83. Levitt, H. (1971) Transformed up-down method in psychophysics. J. Acoust. Soc. Am., 49: 467-477. McFarland, D.J. and Cacace, A.T. (in press) Aspects of short-term acoustic recognition memory: modality and serial position effects. Audiology. McFarland, D.J., Cacace, A.T. and Emrich, J. (1991) Disassociation of auditory and visual short-term memory deficits following temporal lobe damage (abst.), 3rd World Congress of Neuroscience, Montreal, Canada, 180: 26-33. Mesulam, M.-M. (1981) A cortical network for directed attention and unilateral neglect. Ann. Neurol., 10: 309-325. Moore, E.J., Rakerd, B., Robb, R.C., Semela, J.J.M. and Hooks, Jr., W.C. (1989) A microcomputer-based system for auditory evoked potentials. Scand. Audiol., 18: 135141. Musiek, F.E., Baran, J.A. and Pinheiro, M.L. (1990) Duration pattern recognition in normal subjects and patients with cerebral and cochlear lesions. Audiology, 29: 304-313.
Parks, T.E. (1966) Signal-detectability theory of recognitionmemory performance. Psychol. Rev., 73: 44-58. Ringo, J.L. (1988) Seemingly discrepant data from hippocampized macaques are reconciled by detectability analysis. Behav. Neurosci., 102: 173-177. Ringo, J.L. (1991) Memory decays at the same rate in macaques with and without brain lesions when expressed in d' or arcsine terms. Behav. Brain Res., 42: 123-134. Squire, L.R. and Zola-Morgan, S. (1985) The neuropsychology of memory: new links between humans and experimental animals. In: D.S. Olton and S. Corkin (Eds), Memory Dysfunctions: An integration of Animal and Human Research from Preclinical and Clinical Perspectives, Ann. NY Acad. Sci., 444: 137-149. Squire, L.R. and Zola-Morgan, S. (1991) The medial temporal lobe memory system. Science, 253: 1380-1386. Underwood, B.J. (1966) Experimental Psychology, 2nd edn., Appleton-Century-Crofts, New York, pp. 542-580. Warrington, E.K., Louge, V. and Pratt, R.T.C. (1971) The anatomical localization of selective impairment of auditory verbal short-term memory. Neuropsychologia, 9: 377-387. Zwislocki, J., Maire, F., Feldman, A.S. and Rubin, H. (1958) On the effect of practice and motivation on the threshold of audibility. J. Acoust. Soc. Am., 30: 254-262.