J Am Acad Audiol 11 : 273-282 (2000)

Toward a More Ecologically Valid Measure of Speech Understanding in Background Noise James Jerger* Ralf Greenwald* Ilse Wambacq* Amanda Seipel* Deborah Moncrieff,

Abstract

In an attempt to develop a more ecologically valid measure of speech understanding in a

background of competing speech, we constructed a quasidichotic procedure based on the

monitoring of continuous speech from loudspeakers placed directly to the listener's right and

left sides . The listener responded to the presence of incongruous or anomalous words imbed-

ded within the context of two children's fairy tales . Attention was directed either to the right

or to the left side in blocks of 25 utterances . Within each block, there were target (anomalous) and nontarget (nonanomalous) words. Responses to target words were analyzed separately

for attend-right and attend-left conditions . Our purpose was twofold : (1) to evaluate the fea-

sibility of such an approach for obtaining electrophysiologic performance measures in the

sound field and (2) to gather normative interaural symmetry data for the new technique in

young adults with normal hearing . Event-related potentials to target and nontarget words at

30 electrode sites were obtained in 20 right-handed young adults with normal hearing .

Waveforms and associated topographic maps were characterized by a slight negativity in the

region of 400 msec (N400) and robust positivity in the region of 900 msec (P900) . Norms for

interaural symmetry of the P9oo event-related potential in young adults were derived .

Key Words: Dichotic, event-related potential, speech competition

Abbreviations: CVC=consonant-vowel-consonant, EEG =electroencephalic, ERP=event-related potential, ISI = interaural symmetry index, VEOG = vertical electro-oculography

0

ne of the most frequent complaints of hearing-impaired persons is difficulty in understanding a talker when there

is background noise . Competition, either from other talkers or from the common noises of the

environment, is frequently cited as one of the most important factors limiting the usefulness

of hearing aids and other amplification sys-

tems . It is critical, therefore, that the audiolo-gist be able to evaluate the extent to which

background competition limits how much a par-ticular hearing-impaired person can benefit

from amplification .

*School of Human Development, Program in Cognitive

and Neurosciences, and the Callier Center for Communica-tive Disorders, The University of Texas at Dallas, Richardson,

Texas, 'Department of Communication Sciences and

Disorders, University of Florida Reprint requests : James Jerger, 2612 Prairie Creek

Dr . East, Richardson, TX 75080-2679

The importance of testing hearing aid per-

formance in the presence of background noise was recognized more than half a century ago. The

testing protocols developed by Carhart (1946), for example, included word recognition scores in

the presence of sawtooth noise . Subsequent

attempts to devise more sophisticated

approaches did not, however, enjoy widespread

acceptance . The vast majority of audiologists have tended, over the past 50 years, to evaluate

hearing aid performance via word recognition scores in the condition of quiet background . And it is fair to say that, among both researchers and

clinicians, support for this approach has been in

steady decline for many years .

In addition to the oft-cited problems of reli-

ability and significance of differences among aided and unaided listening conditions, clini-cians have questioned the ecologic validity of single-syllable word recognition as a measure of the ability to follow ongoing speech in real-world situations. Ecologic validity is concerned with the

273

Journal of the American Academy of Audiology/Volume 11, Number 5, May 2000

extent to which performance in everyday life can be predicted from test scores (cf. Sbordone, 1997). To be sure, this was certainly Carhart's original goal, and it has always been the goal of thoughtful clinicians . But the approach was based on the assumption that procedures of interest in the laboratory (e.g., the recognition of nonsense syllables and consonant-vowel-consonant [CVC] words) could be extrapolated to the understanding of ongoing speech in a hostile acoustic environment. Modern concepts of ecologic validity stress, instead, the idea of "sampling widely the environments within which a particular task is embedded" (Cole et al, 1997, p. 50) to determine how they affect a subject's responses. We, as audiologists, have adopted a task (e.g ., the recognition of CVC words in quiet) and have assumed that it bears some predictive relation to real-life listening. From the stand-point of ecologic validity, however, it would be more appropriate to sample the spectrum of real-life listening situations and construct tasks more relevant to the various listening chal-lenges represented (for a recent example of this approach, see Flynn and Dowell, 1999).

Of related interest is the development, in recent years, of electrophysiologic measures per-mitting more sophisticated analyses of speech processing (Hall, 1992) Auditory event-related potentials (ERPs) now permit us to evaluate the precise temporal characteristics of the brain's electrophysiologic response to paradigms rang-ing from simple syllable and word recognition (Rugg, 1984 ; Kraus et al, 1992) to the recogni-tion of semantic, syntactic, and morphologic anomalies (Neville, 1985 ; Kutas and Hilyard, 1980; Herning et al, 1987; Osterhout, 1992; Friederici et al, 1993).

With these considerations in mind, we have sought to devise, in our laboratory, a new, more ecologically valid, measure of speech under-standing with the following characteristics:

1. Test stimuli occur in samples of ongoing speech rather than in isolation,

2. Performance is always measured in the presence of background competition, and

3. Performance can be measured electrophysio-logically.

To this end, we have devised a procedure pat-terned loosely after a vigilance task. The speech material is a recording of two classic children's fairy tales, "Cinderella" and "Snow White." The recording is presented via loudspeakers situ-ated directly to the right and directly to the left

of the subject's head. The presentation is qua-sidichotic . The subject hears one part of the fairy tale from one loudspeaker but a different part of the same fairy tale from the other loud-speaker. Throughout the story, some words have been replaced by inappropriate or anomalous words-for example, "Thank you kind fairy god-mother" said Cinderella jump sincerely.. ."

The subject's task is to make an appropri-ate response whenever an anomalous word is heard. The subject is told to respond only to what is heard from one side and to disregard what is heard from the other side . On one half of the trials, attention is directed to the right side, on the other half to the left side . In effect, the subject must attend to what is heard from one direction while simultaneously suppressing what is heard from the other direction and, upon com-mand, to reverse the side to be attended and the side to be suppressed . We have sought, in this paradigm, to create a laboratory version of the situation that hearing-impaired persons fre-quently encounter in real life, the problem of attending to real speech coming from one direc-tion in the presence of competing real speech coming from a different direction.

The purpose of the present study was twofold: (1) to examine the feasibility of such an approach for obtaining electrophysiologic per-formance measures in the sound field and (2) to gather normative interaural symmetry data for the new technique in young adults with nor-mal hearing.

METHOD

Test Stimuli

The subject heard the story of "Cinderella" and a portion of the story of "Snow White," recorded by a male voice and divided into 200 segments at natural pause breaks . Segments were approximately equivalent in length. One hundred of the segments contained target word anomalies. The 100 remaining segments con-tained no anomalies. The segments to be altered were chosen at random . In each of the segments not altered, a single word was randomly chosen as the "nontarget" word for that segment.The test stimuli were stored in digital form and delivered to the subject via digital-to-analog conversion .

Target words were drawn from a pool of 99 primarily single syllable words from the PB lists of Egan (1948). Target words were intentionally selected without regard for (1) the particular semantic and/or morphosyntatic anomaly created

274

Ecologically Valid Measure of Speech Understanding in Background Noise/Jerger et al

or (2) its position in the phrase or sentence . Our

goal was not to study a particular form of anom-aly but simply to randomly interrupt the seman-

tic and/or grammatic integrity of the story.

However, anomalous nouns in the final posi-tion of the sentence or phrase predominated .

Of the total pool of target words, 66 were nouns, 15 were verbs, 12 were adjectives, 5 were adverbs, and 1 was a pronoun . Fifty-seven tar-get words occurred as the final word in the sen-tence or phrase, 42 within the sentence or phrase. Thirty-three target words produced purely semantic anomalies (e.g., "she had to sweep the

sky") . The remaining 66 target words produced

both semantic and grammatical anomalies (e.g.,

"and you shall deep to the ball") . Segments were paired in such a way that the

segment presented from one loudspeaker was displaced by 20 segments from the segment simultaneously presented from the other loud-

speaker . For example, if the segment "Once upon a time" was presented from the right side,

the segment "be allowed to go to the ball" was

simultaneously presented from the left side .

Thus, the subject heard the same fairy tale from

both sides, but the story from one side was

delayed by 20 segments, or about 80 seconds, rel-

ative to the story from the other side . The entire

sequence of 200 segment pairs was divided into 8 blocks of 25 segment pairs each . Within each block there were 12 to 14 target stimuli and 11 to 13 nontarget stimuli . Prior to each block of 25

stimuli, the subject was instructed, via com-

puter monitor screen, to attend either to the

right side and ignore the left side or to attend to the left side and ignore the right side . The

sequence of the 8 blocks was quasirandomly

chosen with the constraint that there must be

exactly 4 listen-right and 4 listen-left condi-

tions . The subject's task was to maintain a silent

count of the number of anomalous words heard

during the block, to report that number at the

end of the block, and then to start a new count

in the next block . Figure 1 illustrates the sequence of target

and nontarget words in each of the two channels in a single block of 25 segment pairs . If the sub-ject was instructed to attend to the right loud-speaker during this block, then the responses to

"T" words represented the "target " condition and

the responses to n words represented the "non-target" condition. Responses to words heard in the nonattended left channel were to be ignored.

The arrows along the lower border of the left loudspeaker sequence, in Figure 1, indicate n seg-

ments during which there was an anomalous,

Right Loudspeaker 1 2 3 4 5 6 7 8 18 19 20 21 22 23 24 25

n T T n T n n n n T n T n T T ; 'n

n n T n n n T T n n T T n T T T

Left Loudspeaker 11 1 I Figure 1 Sequence of target (T) and nontarget (n)

words in a typical block of 25 segments . Sequence is for

the attend-right condition. Arrows indicate segments in

which an anomalous word occurs on the nonattend side

during the presentation of a nontarget word on the attend

side .

potentially intrusive word from the unattended side (i .e ., a target that must be ignored) . By actual count, there were 22 such potentially intrusive words in the nontarget-left and 24 in

the nontarget-right conditions .

Start Ear

Previous research (cf. Boliek et al, 1988) has suggested that, in the directed attention mode, there is an advantage to the first hemi-space to which attention is directed. For this rea-son, the first side attended was randomly assigned to the right side in half of the subjects and the left side in the other half.

Apparatus

The subject was comfortably seated in a dental examining chair. Target stimuli were delivered from two loudspeakers at ear height and 1 meter to the left and right of the subject's ears . A 38-cm computer screen, placed at eye level 1.7 meters in front of the subject, presented directions pertaining to the procedure. The arrangement of the apparatus is illustrated in Figure 2.

Subjects

Twenty young adults (6 men and 14 women) in the age range from 18 to 33 years were tested in two groups of 10 subjects each. All subjects (a) were right-handed, (b) spoke English as a first

language, (c) had normal hearing sensitivity, and (d) had no significant auditory complaint. Subjects were university students who partici-

pated in the project as partial requirement for course credit . Hearing sensitivity was screened on both ears at 20 dB HL from 500 Hz to 4 kHz

275

Journal of the American Academy of Audiology/Volume 11, Number 5, May 2000

Left Loudspeaker

Computer Monitor

1.7 meters

Right Loudspeaker

"Once upon e4 }-N be allowed to a time" go to the ball"

1 meter-----0----*- 1 meter

Figure 2 Arrangement of experimental apparatus. Different parts of the same story are presented simul-taneously from the two loudspeakers .

in octave steps. A response at 20 dB HL at all test frequencies was required for admission to the study.

Recording Apparatus

Brain electrical activity was recorded from 30 silver-silver chloride electrodes mounted in an elastic cap (Neurosoft) and affixed to the scalp according to a modification of the Inter-national 10-20 system (Jasper, 1958). Vertical electro-oculographic activity (VEOG) was mon-itored by electrodes mounted above and below the left eye. Electrical activity was referred to linked mastoid electrodes with Fpz as ground .

Individual sweeps of electroencephalic (EEG) activity, time-locked to the onsets of the target and nontarget words, were stored for offline analysis . Ongoing EEG activity was amplified and bandpass, analog filtered from 0.15 to 125 Hz. Analog filter skirts were sloped at -12 dB per octave . The amplified and filtered EEG activity was then digitized through an acquisi-tion interface system (Neuroscan) and, finally, routed to a computer for individual sweep stor-age, display, averaging, and final digital filter-ing (SCAN 4.0 software). Signal averaging was carried out after offline artifact rejection and baseline correction . Individual epochs were examined and rejected whenever electrical activ-ity in the eye channel (VEOG) exceeded ±50 microvolts . If more than 30 of the 50 epochs of a given target or nontarget condition were rejected for any reason, then the entire data set for that subject was rejected . Thus, each aver-aged ERP was based on a minimum of 20 epochs. Successfully averaged evoked potential wave-

forms were then linearly detrended, digitally lowpass filtered at 20 Hz, and baseline cor-rected . Digital filter slope was -48 dB per octave .

Topographic maps of the digitally filtered auditory ERPs were constructed by interpolation of voltages between adjacent electrodes, using a 4-point linear-interpolation algorithm. Colors representing the range of observed voltages were then assigned to the resulting voltage matrix .

Procedure

Prior to the actual testing of each subject, the speech levels from the two loudspeakers were adjusted to achieve median plane local-ization for that subject. The phrase "once upon a time" was presented diotically, and repeated three times, at each of 11 interaural intensity ratios . The speech level was fixed at 40 dB HL from both loudspeakers . Then the intensity level from the left loudspeaker was varied over a range from +14 to -14 dB in 2-dB steps. After each presentation, the subject judged the appar-ent position of the sound image on an 11-point scale ranging from 5 on the extreme right, through 0 at the midline, to 5 on the extreme left. The interaural intensity ratio corresponding to perception in the median plane was defined as the midpoint of the range of "0" judgments between consistently right- and consistently left-sided responses. The purpose of this proce-dure was to ensure that neither ear enjoyed an intensity advantage as a result of minor inter-aural asymmetries in sensitivity.

After the electrode cap was affixed, the sub-ject was familiarized with the task . Subjects were instructed to count, silently, the number of anomalous words in a given block and to report that number at the end of the block. The purpose of this procedure was to ensure that target words were identified from either side with an accuracy of at least 90 percent. Two practice blocks (25 seg-ment pairs per block) of the fairy tales were played and the subject was instructed to note and count the anomalous words. During actual test-ing, subjects began the silent count anew at the start of each new block. Actual testing began in the "listen-right" condition in half of the subjects and in the "listen-left" condition in the other half. The test phase lasted approximately 25 min-utes with a rest period of about 1 minute halfway through the 8 blocks . Post-testing analysis of the subjects' counts indicated that count accuracy always exceeded 90 percent from both right and left sides.

276

Ecologically Valid Measure of Speech Understanding in Background Noise/Jerger et al

Subjects were tested in two groups . In group A, a 3-second silent interval was inserted between successive segment pairs of the story. In group B, however, no silent interval was inserted . The segment pairs were presented as they had been recorded . The purpose of this grouping was to determine whether the pre-sentation of the story without silent intervals (group B) was too rapid for normal subjects . We hypothesized that a too rapid presentation rate might significantly reduce the amplitude and increase the latency of the response to anomalous words .

-Target Right - - - Target Left - VEOG

10 s s a 2 0

S -2 3 -a d a n

E

Latency (msec)

Data Analysis

Epochs were separately averaged for (1) tar-

gets from the right side, (2) nontargets from the

right side, (3) targets from the left side, and

(4) nontargets from the left side . We had originally

anticipated that there would be evoked potential

activity in response to target words but not to non-target words . The latter words were originally

viewed as a source of baseline activity. Inspection

of grand-averaged waveforms revealed, however,

that, even in the nontarget conditions, there was

slight, apparently evoked, activity. We have

termed this apparent ERP activity the "intru-

sive response," suggesting that it represents the

subject's inability to totally suppress target words

from the not-to-be-attended side (see Figure 1) .

Interesting though these grand-averaged responses might be, their amplitudes in individ-

ual subjects were only slightly above the back-

ground noise level . Statistical analysis was

confined, therefore, to the two target conditions .

The mean amplitudes and latencies measured at

the peak positivity of the individual ERP wave-

forms at electrode CPZ in the target-right and

target-left conditions were compared as a func-

tion of group (A or B) and target side attended

(right or left) by means of conventional Student's

t-tests . Statistical significance was evaluated at

an alpha level of 0.05 .

RESULTS

Illustrative Waveforms

Figure 3 shows illustrative averaged wave-

forms from two subjects at electrode CPZ. For each subject, three waveforms are shown: (1) target right, (2) target left, and (3) VEOG. These wave-

forms are representative of the entire group .

The principal finding was a robust positivity in the latency region from 700 to 1100 msec . The

Figure 3 Illustrative averaged waveforms from 2 of the

20 subjects . Waveforms are shown for the target-right and

target-left conditions at electrode CPZ, and for the verti-

cal eye channel (VEOG) .

region of positivity was centered in the midline centroparietal region of the scalp . Further,

responses to targets presented from right and left sides were approximately equivalent .

Statistical Analysis

In order to determine whether the variables

of group (A or B) or side attended (target right

vs target left) exerted a significant influence,

individual averaged waveforms were analyzed

for amplitude and latency in the 700 to 1100 msec

latency region . Table 1 summarizes means and

standard deviations of these data . The amplitude

tended to be slightly greater in the target-left

than in the target-right condition, and latency

was slightly longer in the target-left condition,

but Student's t-tests, comparing groups and side

attended, failed to reveal any statistically sig-

nificant mean differences . It is particularly note-worthy, moreover, that neither amplitude nor

latency suffered in group B, where segments

were delivered without pauses between seg-

ment pairs . These results argue that the 3-

second silent intervals between segment pairs

were not necessary in order to ensure that the

pace of the listening task was not too rapid for

young adults with normal hearing . In view of the nonsignificant findings, data

were collapsed across group and side-attended for subsequent analysis .

Topographic Distributions of ERP Activity

Figure 4 shows the waveform, grand-averaged across all 20 subjects and across both

277

Journal of the American Academy of Audiology/Volume 11, Number 5, May 2000

Table 1 Means (and SDs) of Amplitudes and Latencies at the Positive Peak of the Waveform at Electrode CPZ for Target-right and Target-left

Conditions in Two Groups of Young Adults with Normal Hearing

Group n Target

Amplitude (AV)

Right

Latency (msec)

Target

Amplitude (p, V)

Left

Latency (msec)

A* 10 4.39(l .47) 849.2 (69 .3) 4.83(l .97) 884.6 (130 .2) Bt 10 6.41 (2 .66) 841 .8 (107 .5) 6.18 (2 .27) 913 .3 (95 .7) Total 20 5 .40 845.5 5.51 899.0

*Three-second silent intervals between segments; tno silent interval between segments .

target-right and target-left conditions, at each of the 30 electrode sites, along with the wave-form on the vertical eye channel (VEOG) . In the frontal region, especially at electrodes F7, FT7, F8, and FT8, there is a negative peak at about 400 msec . In the midline, however, this nega-tivity is attenuated by the strong positivity arising from the centroparietal region between 400 and 500 msec . This interaction is high-lighted in Figure 5 . Here, serial topographic brain maps are shown at 25-msec intervals over the latency range from 200 to 500 msec . Negativity begins to build in the frontal region at 240 msec, but strong positivity centered in the midline centroparietal region begins shortly after 400 msec and quickly swamps the frontal negativity.

This interaction between frontal negativ-ity in the 200- to 500-msec region and cen-troparietal positivity in the 500- to 1100-msec region is further illustrated in Figure 6. Here, the waveforms at electrodes F, F4, F3, F8, and F7 are shown. At the two most lateral electrodes (FS and F7), one can clearly see the classic N4oo response to semantic anomaly. The negativity begins in the 200-msec region, peaks between 500 and 600 msec, and returns to baseline in the 900- to 1000-msec region . But at the more cen-tral electrodes (FZ, F3, and F4), the initial negativity is reversed by the onset of the midline cen-troparietal positivity, which begins at about 400 msec and peaks at about 900 msec .

The serial topographic maps of Figure 7, executed at 50-msec intervals over the latency

12laV E

Figure 4 Grand-averaged wave-forms at each of the 30 scalp electrode sites and the vertical eye channel. Average waveforms have been col-lapsed across group and side attended.

278

Journal of the American Academy of Audiology/Volume 11, Number 5, May 2000

semantic and syntactic, it is not surprising that we observed only a slight negativity in the 400-msec region (associated with a purely seman-tic anomaly), followed by a pronounced positiv-ity in the 600- to 1000-msec region (associated with syntactic anomaly) . Whether our positiv-ity was generated by the grammatic violations or simply by the unexpectedness of the anom-alous words is not easily resolved. Our purpose, however, was not to study language processing per se, but only to study interaural asymmetry in the processing of anomalous target words presented in a quasidichotic fashion in the sound field .

The present dichotic listening paradigm is, to a certain extent, unique in the sense that it probes the ability to detect semantic and mor-phosyntactic violations in relatively ongoing speech heard from one side in the presence of competing speech heard from the other side . A possible clinical application might be the eval-uation of performance with and without ampli-fication . Within the paradigm, one might compare unaided with aided performance and/or compare various aided conditions (e.g., mon-aural vs binaural systems or conventional aid vs assistive listening device). The procedure could also be used clinically to evaluate interaural asymmetry (e.g ., in individuals with suspected central auditory processing disorder).

of semantic, morphological and syntactic violations . Cognitiue Brain Research 1:183-192 .

Gunter T, Stowe L, Mulder G. (1997) . When syntax meets semantics. Psychophysiology 34:660-676 .

Hall J . (1992) . Handbook ofAuditory Euoked Responses . Boston : Allyn & Bacon .

Hansen JC, Hillyard SA. (1980) . Endogenous brain poten-tials associated with selective auditory attention. EEG Clin Neurophysiol 49 : 277-290.

Herning R, Jones R, Hunt J. (1987) . Speech event related potentials reflect linguistic content and processing level. Brain Lang 30:116-129 .

Jasper H. (1958) . The ten-twenty electode system of the International Federation . EEG Clin Neurophysiol 10:371-375 .

Juottonen K, Revonsuo A, Lang H. (1996) . Dissimilar age influences on two ERP waveforms (LPC and N400) reflect-ing semantic context effect . Cognitiue Brain Research 4:99-107.

Kraus N, McGee T, SharmaA, Carrell T, Nicol T. (1992) . Mismatch negativity event-related potential to speech stimuli. Ear Hear 13:158-164.

Kutas M, Hilyard S. (1980) . Reading senseless sentences: brain potentials reflect semantic incongruity. Science 207:203-205 .

McCallum W, Farmer S, Pocock P. (1984) . The effects of physical and semantic incongruities on auditory event-related potentials . EEG Clin Neurophys 59:477-488 .

Naantanen R. (1982) . Processing negativity : an evoked potential reflection of selective attention. Psychol Bull 92 :605-640 .

REFERENCES

Boliek C Obrzut J, Shaw D. (1988) . The effect of hemis-patial and asymmetrically focused attention on dichotic listening with normal and learning-disabled children . Neuropsychologia 26 :417-433 .

Carhart R. (1946) . Tests for the selection of hearing aids . Laryngoscope 56:780-794.

Cole M, Hood L, McDermott R. (1997) . Concepts of eco-logical validity : their differing implications for comparative cognitive research . In : Cole M, Engstrom Y, Vasquez O, eds . Mind, Culture, and Activity . Cambridge, UK: Cambridge University Press, 49-56.

Coulson S, King J, Kutas M. (1998) . Expect the unex-pected : event-related brain response to morphosyntactic violations . Language & Cognitiue Processes 13:21-58 .

Egan J. (1948) . Articulation testing methods . Laryngoscope 58:955-991 .

Flynn M, Dowell R. (1999) . Speech perception in a com-municative context: an investigation using question/ answer pairs . J Speech Lang Hear Res 42:540-552 .

Friederici A, Pfeifer E, Hahne A. (1993) . Event-related brain potentials during natural speech processing: effects

Neville H. (1985) . Brain potentials reflect meaning in language . Trends Neurosci 8:91-92 .

O'Halloran JP, Isenhart R, Sandman CA, Larkey LS . (1988). Brain responses to semantic anomaly in natural, continuous speech . Int J Psychopshysiol 6:243-254 .

Osterhout L. (1992) . Event-related potentials elicited by syntactic anomaly. J Mem Lang 31:785-806 .

Osterhout L, Hagoort P (1999) . A superficial resemblance does not necessarily mean you are part of the family : counterarguments to Coulson, King and Kutas (1998) in the P600/SPS-P300 debate . Language & Cognitive Processes 14:1-14.

Osterhout L, Holcomb P. (1993) . Event-related potentials and syntactic anomaly: evidence of anomaly detection during the perception of continuous speech . Language & Cognitiue Processes 8:413-437 .

Rugg M. (1984) . Event-related potentials and the phono-logical processing of words and nonwords . Neuropsycho-logia 22:435-443.

Sbordone R. (1997) . The ecological validity of neuropsy-chological testing. In : Horton A, Wedding D, Webster J, eds. The Neuropsychology Handbook. Vol 1. 2nd Ed . New York : Springer, 365-392.