Perception & Psychophysics

1974, Vol. 16, No.1, 143-149

Effects of noise letters upon the identification of a

target letter in a nonsearch task*

BARBARAA. ERIKSEN and CHARLES W. ERIKSEN

University ofIllinois, Champaign/Urbana, Illinois 61820

During a I-sec tachistoscopic exposure, Ss responded with a right or left leverpress to a single target
letter from the sets Hand K or Sand C. The target always appeared directly above the fixation cross.
Experimentally varied were the types of noise letters (response compatible or incompatible) flanking the
target and the spacing between the letters in the display. In all noise conditions, reaction time (RT)
decreased as between-letter spacing increased. However, noise letters of the opposite response set were
found to impair RT significantly more than same response set noise, while mixed noise letters belonging
to neither set but having set-related features produced intermediate impairment. Differences between
two target-alone control conditions, one presented intermixed with noise-condition trials and one
presented separately in blocks, gave evidence of a preparatory set on the part of Ss to inhibit responses
to the noise letters. It was concluded that S cannot prevent processing of noise letters occurring within
about 1 deg of the target due to the nature of processing channel capacity and must inhibit his response
until he is able to discriminate exactly which letter is in the target position. This discrimination is more
difficult and time consuming at closer spacings, and inhibition is more difficult when noise letters
indicate the opposite response from the target.

A popular experimental approach to the study of target identification, we need an anchor or baseline
visual information processing has been to employ a condition, i.e., what is the effect of noise on speed or
display in which a target is embedded among a number accuracy in target identification when no visual search is
of noise elements. Typically, the target has been a letter required? Can a letter that always appears in the same
embedded in a display containing other letters. Initially, known location in the visual field be identified as rapidly
a major concern of this research was whether the search when surrounded by noise letters as when it appears
process itself was serial or parallel in nature. However, it alone? The search task by definition requires some
soon became apparent that many important questions processing of noise in order to locate the target. Thus,
centered around the effects of the noise stimuli. inferences as to the effects of noise are confounded with
As would be expected, accuracy or speed in assumptions as to how this search of the display is
identifying the target was found to be dependent upon carried out.
the similarity of the target and noise (McIntyre, Fox, & There is strong evidence that even when search is
Neale, 1970; Estes, 1972). Reaction time (RT) to the eliminated, noise letters or other stimuli impair the
target increased, and probability of its detection processing of the designated target letter. Eriksen and his
decreased as the number of noise elements increased associates (Eriksen & Hoffman, 1972, 1973; Colegate,
(Estes, 1972). In attempts to understand how these Hoffman, & Eriksen, 1973) have employed circular
noise variables entered into the search task, it became displays in which an indicator line is presented to the S
necessary to control for such variables as the retinal at various intervals before the target letter and the
locus of the noise and the target and the spacing remainder of the display appears. The indicator clearly
between the target and the closest noise letters. designates to the S the location of the target letter.
Questions were raised as to the extent to which noise While this research has been focused upon the nature of
letters were processed, and the possibility of shared selective attention that is involved, there have been three
feature analyzers or detectors was considered. major findings that are directly relevant to the visual
In spite of repeated investigations of the above search task and the effects of noise upon target location
parameters, the nature of the search process and the and identification. First, attentional selectivity is unable
effects of noise elements is far from resolved (Estes, to eliminate completely the effects of extraneous
1972; Kinchla, 1974; Gardner, 1973; Shiffrin & stimuli. Even when the S is provided with an indicator as
Gardner, 1972). At this point in the research effort it much as 400-500 msec before the target and the noise
would appear fruitful to modify the experimental task. elements appear (ample time to process the indicator
To understand the search task and the effect of noise on information), the presence of noise letters in the display
delays RT to the target letter. Second, the spatial
'This investigation was supported by U.S. Public Health proximity of noise letters to the target has a nonlinear
Service Research Grant MH-1206 and U.S. Public Health Service
Research Career Program Award K6-MH·22014. effect upon target RT. Noise letters within ~ deg of

143
144 ERIKSEN AND ERIKSEN

Table 1 are not detectable in terms of further reductions in RT.

Experimental Conditions and Representative Displays In terms of the spatial proximity effect found with
Condition Example circular displays, a separation between noise letters and
target of greater than approximately ~ deg produces
1 Noise Same as Target H H H H H H H
2 Noise Response Compatible K K K H K K K little or no further decrease in RT.
3 Noise Response Incompatible S S S H S S S To recapitulate, the S in an experiment using circular
4 Noise Heterogeneous-Similar N W Z H N W Z displays with a leading indicator experiences an indicator
5 Noise Heterogeneous-Dissimilar G J Q H G J Q which directs attention to a particular position in the
6 Target Alone H
display. However, the minimal capacity of his visual
attentional channel is such that he will process two or
visual angle of the target produce a much greater more letters from that sector of the display. He thus has
impairment in RT than those further removed. In fact, available essentially simultaneously from the processing
once the noise elements are separated by a degree of side of the chain several letters that start to evoke or
visual angle from the target, further separation appears carry through as responses. An inhibitory process is
to make little if any difference. Third, the effect of noise required to prevent the responses from running off
letters on target RT is predominantly on the response willy-nilly. During this inhibitory period, a selection
side as opposed to the processing side. If noise letters must be made from the various alternative stimuli that
have the same prior learned response as the target letter have been processed. In the case of the indicator display,
but are physically quite different, little impairment in this selection is made in terms of determining which of
RT to the target letter is obtained (Eriksen & Hoffman, the available letters matches the indicated position on
1973). On the other hand, if the noise letters require a the display. This selection is more rapid if the spatial
response opposite or incompatible to that of the target discrimination is easy.
letter, a large impairment in RT is obtained. This finding
would tend to place the locus of the effect in terms of
METHOD
response competition.
The above findings suggest that visual attention is not Design and Rationale
capable of. infinitely fine selectivity. Rather, there is a In the present experiment, visual search was eliminated by
minimal channel size in terms of capacity for having the target letter always appear in the same location,
which was Yz deg above the fixation point. S was thoroughly
simultaneous processing with a capacity exceeding that instructed to respond only to the letter in this location and to
required for identifying a single letter. Further, the ignore any and all other letters that might appear simultaneously
unutilized capacity cannot be shut off and, if there are with the target. The S was required to press a lever to the right
other letters or stimuli present, they will be processed (or left) if the target letter was an H or a K and in the opposite
simultaneously along with the target. In other words, the direction for an S or a C. Except for two control conditions in
which the target letter appeared alone in the display, the target
channel is filled up to its minimal capacity if material is was always flanked on either side by three noise letters. There
present to fill it. The result is that the S has available 2, were five conditions of noise: (1) noise identical to target-the
3, or perhaps 4 or more letters that complete processing target letter was flanked on either side by three repetitions of
at essentially the same time and begin the process of the target letter; (2) noise same response as target-the target
letter was flanked on either side by three repetitions of the letter
evoking responses. that was the other member of that target set; (3) noise other
But at this point the information processing system target set-target flanked by three occurrences of one of the
experiences a severe limitation. Only one letter at a time letters of the opposite response set; (4) noise similar to
can be vocalized overtly or covertly, and a lever cannot target-target flanked by letters having features similar to the
be moved both right and left at the same time. target set as determined by the Gibson system (Gibson, 1969)
but excluding the two letters that were members of the target
Therefore, a selection process is required to determine set; (5) noise dissimilar to target-target flanked by letters having
which stimulus will evoke a response and which stimuli features dissimilar to target set as determined by the Gibson
will be inhibited. This selection and inhibition requires a system and excluding the target set. In Table 1, the conditions
finite period of time which adds to the RT in target are defined with the examples of the corresponding displays in
which the target letter in all cases is an "H." Table 2 presents the
identification or detection. noise letters used for the similar and dissimilar noise conditions
It is here that the spatial proximity effect can be and compares their similarity to the letters in the two target sets
accounted for. In selecting which of several available in terms of the Gibson system.
letters to emit as a response, the S must perform the A second experimental manipulation concerned the spacing
additional task of determining which one of these letters between the adjacent letters in the display. All five noise
conditions were run under each of three different spacings. In
came from the indicated position. In essence, a spatial the closest spacing, the nearest contours of adjacent letters were
discrimination is required. As is true of discrimination in separated by approximately .06 deg of visual angle, a distance
general, small stimulus differences require longer corresponding to the spacing of letters within a printed word.
decision times. But the RT-stimulus difference Spacings of Yz deg and 1 deg of visual angle were also employed.
During the experimental sessions, data were collected for two
magnitude function becomes asymptotic rather quickly control measures. Both involved RT to target letters when no
as the stimulus differences increase. Once stimulus noise letters were present in the display. In the target-alone
differences achieve some minimal value, further increases mixed control, the single-letter display was presented during
 EFFECTS OF NOISE LETTERS 145

blocks of trials in which all five experimental conditions were Table 2

represented. In the single-letter blocked control, RTs to the Letter Features as Rated by the Gibson System*
target alone were obtained with the single-letter displays
presented in blocks of 12 trials, during which the S knew that H & K (NWZ) S&C(GJQ)
the displays containing noise elements would not appear. Features H K N W Z S C G J Q
In terms of our analysis of the processes involved in this
experimental task, we anticipated that faster RTs would be Straight
obtained using the blocked trials. When single-letter displays are Horizontal + + +
randomly intermixed with noise letter displays, the S will still be Vertical + + + "
prepared to inhibit responses to noise letters that are processed Diagonal / + + +
along with the target. This inhibition must continue until S Diagonal \ + + + +
verifies which letter occurred in the target position. In the Curve
blocked trials, S can abandon the inhibitory process, since no Closed +
noise letters are present and no position verification is required. Open V +
It is also possible that if letter identification occurs by OpenH + + + +
subanalysis or subprocessing of features, then 5 could reduce his Intersection + + +
criterion for a response, basing it upon fewer features. Redundancy
With respect to the experimental conditions, we anticipated Cyclic Change + +
that the longest RTs would be obtained for the condition in Symmetry + + +
which noise letters consisted of repetition of a letter from the Discontinuity
opposite target set. If one or more of the noise letters were Vertical + + +
processed simultaneously with the target, then competing Horizontal +
responses would tend to be evoked, requiring not only inhibition Totals 5 5 3 4 3 2 2 2 2 3
but maximum care in position verification. This condition ---
should show rather marked improvement in RT as a spacing 10 7
between the letters becomes greater and therefore the spatial
discrimination easier. Similarly, if the effect of noise letters is "Gibson, 1969, p. 88
primarily due to competition for a single response channel, then
minimal effects of noise should be obtained for the identical and right or left upon identification of the target stimulus. Pressing
same response noise conditions. Since in both of these of the lever also stopped the Klockounter. A pair of lights below
conditions noise letters are eliciting the same response as the the Klockounter indicated to E whether 5 had pressed the switch
target letter, we anticipated little or no difference between them. in the correct direction for the trial. RT was recorded in
If a significant portion of the effect of noise on target milliseconds.
identification is attributable to the processing of the stimuli, Target stimuli were the capital letters H, K, S, and C. For the
then differences between the identical and same response noise conditions employing noise letters having angular features, the
might be anticipated. If we think in terms of feature detectors, letters used were N, W, and Z. The curved-feature letters were G,
the identical noise would be competing for the same feature J, and Q. All letters were capitals from Paratipe 18-point Futura
detectors which might be expected to slow the processing of the Bold style. Since Fields 2 and 3 of the tachistoscope were used
target. On the other hand, the same response noise would be for the target and noise letters, respectively, and these fields
using somewhat different feature detectors, although having reverse the image, the letters were placed on transparent vinyl
some in common. Both conditions would require position cards which were inserted backwards into the holders. White
verification on the part of S, since he needs to verify one of the vinyl cards installed in the holders provided the necessary
letters he has available as appearing in the target position. Thus, contrast for the displays.
both conditions should show an improvement in RT as the The cards containing the target letters each had a single letter
spatial position variable becomes more discriminable. which appeared centered V, deg of visual angle above the fixation
The noise-similar and noise-dissimilar conditions are of special cross. On the noise-letter cards, all three noise letters appeared
interest. If letter recognition occurs by a feature-analyzing on either side of the target. Three different orders were used for
process which, in turn, corresponds to Gibson's classification each set so that each letter appeared in each possible position in
system, we might anticipate that the noise-similar condition relationship to the target letter. In addition, noise-letter cards
would produce longer RTs than the dissimilar condition due to were constructed using each target letter repeated six times.
competition with the target for feature analyzers. On the other Three different spacings of the noise letters in relationship to the
hand, a process of stimulus generalization would seem to predict target letter and each other 'were also employed. For the closest
that similar features would facilitate the response to the target spacing, the letters were placed as if printed material were being
letter, since the features of the noise letters would tend to elicit constructed. The average between letter distance for this spacing
the same response as the target. was approximately .0625 deg of visual angle at the viewing
distance of 48 in. Between letter (edge to edge) spacings of
Subjects V, deg and 1 deg of visual angle were employed in the other two
The six Ss (three male) were paid volunteers, all of whom spacing conditions. Fixation was provided by a I/8-deg cross on
were graduate or undergraduate students at the University of a white-backed transparent card in Field I.
Illinois. All had normal or corrected to normal vision and were
right-eye dominant and right-handed. Procedure
Ss were instructed to hold the handswitch for initiating the
Apparatus and Stimuli stimulus presentation in the left hand and to place the right hand
Stimuli were presented in a Scientific Prototype Model GA on the handrest with thumb and forefinger holding the response
three-field tachistoscope in which the original lamps had been lever. They were told to be sure the fixation cross was in good
replaced with Sylvania F4T5CWX bulbs. Luminance of all fields focus before initiating a trial. Three of the Ss were instructed to
was maintained at 10 fL, as measured by a Spectra brightness press the lever to the right if the target letter was an H or a K
spot meter. Onset of the stimulus was initiated by S, using a and left for an S or a C. The other three Ss received the opposite
hand switch in his left hand, which also started a Hunter instructions. The target always appeared directly above the
McidelI522 digital Klockounter. A handrest to the right of S fixation point. Particular care was taken to insure that S
contained a small lever switch which S could press either to the understood that the target always appeared in the same location
 146 ERIKSEN AND ERIKSEN
550 Conditions: The effects for spacing and noise conditions were both
significant beyond the .0001 level with Fs of 58.28, df =
10-0
2e-e
3o-<J 2,10, and F = 28.23, df = 4,20, respectively. The S'
4_
Sir-A
variable was significant at the .05 level. There was also a
significant Noise Condition by Spacing interaction
[F(8,40) = 16.08] .
The nature of the results can be seen in Fig. 1, where
500 mean RT across Ss is plotted as a function of spacing for
each of the five noise conditions. All five noise
conditions showed decreasing RT as separation between
... the display letters increases from .06 to 1 deg of visual
Ql
on angle. The Spacing by Conditions interaction reflects the
E
fact that the difference between conditions is greatest at
the closest spacing. When the display letters have been
450 separated by 1 deg of angle, performance for the
noise-identical, noise-same-response, and noise-similar
conditions is essentially the same. Over all values of the
* control
mixed spacing variable, RT to the condition in which the noise
was a letter of the opposite response set was the longest,
while the fastest RTs were obtained when the noise was
identical to the target. There is little difference between
* control
blocked this latter condition and the condition in which the
.5· r noise letter was the other member of the same target set
Between-Letter Spacing
(calling for the same overt response). The lack of a
significant difference between these two conditions is
Fig. 1. Mean reaction times (R Ts) as a function of spacing (six attested to by the finding that half of the Ss were faster
Ss combined) for the five experimental conditions and two under the noise-same-response condition than they were
control conditions. Experimental conditions are as follows: when the noise was identical to the target.
(1) noise same as target; (2) noise response compatible; (3) noise
response incompatible; (4) noise heterogeneous similar; (5) noise With the exception of the closest spacing, the
heterogeneous dissimilar. dissimilar-noise condition, in which the noise letters
exhibited features similar to letters of the opposite
and that this was the only letter he was to respond to. Viewing target set, gave longer RTs than when the noise letters
was monocular with the right eye. The fixation field went off contained features similar to the target.
and the target and noise letters appeared simultaneously as 8
pressed the handswitch button. They remained on for I sec and When the letters of the display are separated by 1 deg
terminated together, at which time the fixation field came back of visual angle, the noise-similar, noise-identical, and
on. Two small lights, visible only to the E, indicated the noise-same-response conditions yield RTs that are
direction in which the lever had been pressed. Any trials in essentially the same as those obtained for trials in which
which 8 pressed the lever in the wrong direction or both
directions were not recorded and were rerun later in the session.
the target letter alone is presented in blocks of trials
A note was made of the error. containing the other noise conditions. However, this
All combinations of conditions by spacing were randomly mean RT value is appreciably above that obtained for
intermixed within blocks of 32 trials. Four trials in each block the target letter alone in blocked trials. The difference is
contained the target letter alone. An additional block of 12 trials approximately 30 msec. The difference between the two
in the single-letter alone condition was run during each session.
For experimental presentation, Conditions 4 and 5 (see Table 1) target-letter-alone conditions was tested in a three-way
were subdivided as follows: 4a-curved target letters (8 and C) analysis of variance (Ss by Target Letter by Mixed vs
with similar noise (G, J, and Q); 4b-angular target letters (H and Blocked Trial Presentation). The effect for mixed vs
K) with similar noise (N, W, and Z); Sa-curved target letters (8 blocked trials was significant [F(I,5) = 42.38, p < .01],
as was that for Ss [F(5,3) = 20.87, p < .05] . There was
and C) with dissimilar noise (N, W, and Z); and 5b-angular
target letters (H and K) with dissimilar noise (G, J, and Q).
Each S participated in two practice sessions before beginning also a significant Ss by Letter interaction [F(15,15) =
the experiment. Six experimental sessions were run, each 7.59, p < .001], reflecting the fact that all Ss tended to
consisting of three blocks of 32 trials, plus one block of 12 make the lever response to the right somewhat faster.
single-letter trials. This yielded a total of 24 trials per S for each There was no significant effect attributable to RTs to
condition by spacing combination plus 18 trials per 8 for each
single letter in each of the two control conditions (mixed and the different target letters in the response sets.
blocked). The data for the noise-similar and noise-dissimilar
conditions were also analyzed separately in a three-way
RESULTS analysis of variance (Ss by Spacing by Conditions). Both
the spacing and noise variables were significant at or
The RT data were first analyzed in a three-way beyond the .01 level [F(2,10) = 55.32 and F(1,5) =
analysis of variance (Ss by Spacing by Noise Conditions). 15.34, respectively] . The interaction between these two
 EFFECTS OF NOISE LETTERS 147

Table 3
Errors Per Noise Condition: Six Ss Combined
.06 Deg Yz Deg 1 Deg
Number Per- Number Per- Number Per-
Condition of Errors cent of Errors cent of Errors cent
Noise Homogeneous
31 21.5 24 16.6 6 4
Target Opposite Response Set
Noise Homogeneous N=
144 1 <1 4 2.7 6 4
Target Other Member of Same Response Set
Noise = Target 6 4 5 3.4 4 2.7
Noise Heterogeneous
Target Features Opposite Response Set 34 11.8 15 5.2 10 3.5
N=
Noise Heterogeneous 288
Target Features Same Response Set 20 6.9 5 1.7 18 6.3

variables was significant beyond the .05 level [F(2,1O) = possibility of rather gross effects such as these upon
4.83] . input processing can be discarded because the delaying
The response errors were also examined. For the two effect of noise is markedly determined by its response
control conditions, target letter alone-mixed had a 5% compatibility with the target. In terms of physical
error rate as contrasted with a 7% error rate for target characteristics of the stimulus, flanking the target letter
alone in blocked trials. This does not appear to be a by a different letter that has the same learned response
sufficient difference to attribute the superiority of the as the target would seem to provide as much physical
blocked trials to a speed-accuracy tradeoff. In fact, for complexity as flanking the target by a letter requiring an
one S who made only one error on all target-alone mixed opposite response. Yet in only the latter condition is RT
trials and two errors on target-alone blocked trials, there to the target materially increased. Since the effect of
was still a large difference between the two in RT: an noise is strongly determined by its response
average RT of 423 msec on the mixed and 405 msec on compatibility with the target letter, support is given not
the blocked. only to the conclusion that the effects of noise are the
A breakdown of the error data for the different noise result of response competition or interference, but
conditions is presented in Table 3. These error data further that this results from. at least some of the noise
reflect essentially the same differences between noise stimuli being processed along with the target to the
conditions and the effect of the spacing variable as was point where they are identified enough to tend to elicit
obtained for the RT measures. It is readily apparent that appropriate responses. This conclusion is commensurate
noise of the opposite response set not only slowed Ss' with our suggestion, made in the introduction of this
responses but caused them to err much more frequently paper, that there is a limit to the degree to which a
at the two closest spacings. Similarly, noise consisting of human S can restrict his visual attentional capacity and
letters having features similar to those of the other targetthat this minimal level is in excess of the capacity
set tended to have much the same effect. required for processing a singleletter.
To the extent that one can make a distinction
DISCUSSION between the input side of stimulus processing and the
output or response side, we feel that the present data
It is quite clear from the above data that even when favor a theory that putsthe locus of the effect of noise
an S is not required to search for a target letter in a at the stage of response selection. One finding in the
multiletter display, the presence of other letters slows present data that would tend to implicate an effect at
his RT to identification of a target letter. Further, they the input processing stage is the finding that even
exert a retarding effect upon RT even when the precise response-compatible noise (noise-identical and
location of the target letter is known to S and the noise-same-response conditions) gave longer RTs at the
target's position in the visual field is favored over the closest stimulus spacings compared with the no-noise
noise letters in terms of its location on the fovea. Thus, mixed condition. If one adopted a limited-energy
the effect of noise letters upon target identification is parallel processing model (Rumelhart, 1970), the effect
not something that is inherent in the search process. could be attributed to the distribution of the limited
However, the question remains as to how this noise energy over a larger number of stimulus elements with
introduces its effects upon the idenfication process. an attendant reduction of the speed with which any
From the present data, it is apparent that the effect of single element was processed. In this view, even when
noise is not at some gross level. It is not a sort of the noise letters lead to the same response as the target,
"distraction effect," nor a primitive perceptual process their presence in the display would require some of the
that involves the perception of a whole before its parts available processing energy, resulting in the slowing of
can be analyzed, nor a rudimentry noting of the the processing of the target.
presence or absence of items in the visual field. The A decrement resulting from energy limitations would
148 ERIKSEN AND ERIKSEN

have to be in addition to the observed response in terms of letter position requires that Ss use some type
interference effect. Such a model would also have to of inhibitory process in the task to prevent responses
account for the prominent effect we obtained as a from occurring until the selection has been made. The
function of the spacing between target and noise letters. presence of such an inhibitory process would account
It is not clear why processing capacity would have to be for the differences between these two control
shared when the stimuli were .06 deg of angle apart but conditions. In the mixed control, the target letter
not when separated by ~ or 1 deg of visual angle. appeared without noise but on unannounced trials that
Our interpretation of the obtained spacing effect is occurred during blocks in which all experimental
that it represents the ease with which a S can make a conditions were represented. In this case, when S
spatial or location discrimination. If the S is processing, initiated a trial it would be to his advantage to have
essentially simultaneously, the target letter and one or inhibitory processes activated, whereas in the blocked
more noise letters, some form of an inhibitory process condition S knew that all trials in the 12-trial block
has to be activated in order to prevent responses to the would consist only of the target letter being presented.
noise letters and to permit selection of the response RT in that latter condition averaged approximately
appropriate to the target location. This response 30 msec faster than in the other control. Here the S
selection must be made in terms of the target letter's could have discarded any inhibitory process in
location in the display. As with discrimination tasks in preparation for a trial.
general, the more discriminable the differences in An interpretation of noise effects as resulting from
location in the display, the faster will be the selection interference at the response selection stage requires a
process. This pin-pointing of the target's location would closer examination of the results obtained from the
be slowest when the spatial separation was quite small. noise-similar and noise-dissimilar conditions. At the
As the spatial separation increased, the decrease in RT closest spacings, these two conditions gave RTs midway
for this discrimination would rapidly become in the range between same response noise and opposite
asymptotic. Once a difference is big enough, further response noise. At the two wider spacings, similar noise
increases do not speed discrimination RTs. Even on gave RTs that were not reliably different from those
those trials where all the noise responses led to the same obtained to iden tical or same response noise, while
response as the target letter, it is reasonable to expect dissimilar noise approached the RTs obtained under
that the S would still have a check operation to verify opposite response noise condition. That both similar and
that the letter to which he was responding was indeed dissimilar noise have the same effect on RT at the
the one in the target location, since it was necessary for .06-deg spacing is not readily explainable. However, at
him to be prepared on every trial for any of the noise the two wider spacings, their performance is consistent
conditions. with what would be expected in terms of stimulus
Other possible sources of the spacing effect seem less generalization. Similar noise letters had features in
likely. At the closest spacing, the possibility of contour common with the target letter, whereas dissimilar noise
interference between adjacent letters exists (Flom, letters had features in common with the opposite target
Weymouth, & Kahneman, 1963). If present, its relative set. To attribute the effect to stimulus generalization,
contribution would be slight since the spacing variable however, does little more than relate the finding, to an
interacts markedly with the response compatibility of established phenomenon in the experimen tal literature
the noise letters, and there is no obvious reason why (Guttman & Kalish, 1956).
contour interference between letters should be greater if The questions of interest at this point are what
their responses are incompatible than if they are mediates stimulus similarity and how is a common
compatible. response facilitated? Bamber (1969) has proposed an
Since the target letter always appeared ~ deg above "identity detector" based on the physical identity of
the fixation point, noise letters fell on progressively less two stimuli. However, our data indicate that both a
acute areas of the fovea as the spacing between the physical and functional identity are detected with equal
stimuli increased to 1 deg of angle. It is possible that speed.
transmission time for fibers in the retina increases for Estes (1972) and Bjork and Estes (1973) have
receptors farther from the center of the fovea. Further, attributed the effect of noise letters to possible
Sternberg (1967) has shown that degrading a stimulus competition between or an inhibition among feature
leads to an increase in its processing time. While such analyzers. Their interpretation places the effects of noise
effects may possibly be involved in the present spacing on the input side of processing. A suggestion by Bjork
effect, their contribution would be minimal in view of and Estes (1973) that more complex letters with more
the marked interaction between spacing and the features produce more lateral masking and are in turn
response compatibility of the noise letters. more resistant to lateral masking was not borne out by
Our view that the spacing effect primarily reflects the the present experiment. It is, of course, possible that the
ease of response selection is supported by the obtained strong effect of response conflict obscured any effects of
difference between the two control conditions-target feature complexity. While our interpretation emphasizes
alone mixed vs target alone blocked. Response selection the locus of the effect at a response selection stage, it
 EFFECTS OF NOISE LETTERS 149

does not preclude the possibility of submechanisms such resolution and contour interaction. Journal of the Optical
Society of America, 1963,53,1026-1032.
as feature detectors playing a role on input processing .. Gardner, G. T. Evidence for independent parallel channels in
Indeed, such mechanisms as this may be the basis for the tachistoscopic perception. Cognitive Psychology 1973 4,
130-155. ' ,
stimulus generalization effects. Gibson, E. J. Principles of perceptual learning and development.
New York: Appleton-Century-Crofts, 1969.
Guttman, N., & Kalish, H. I. Discriminability and stimulus
REFERENCES generalization. Journal of Experimental Psychology, 1956 51,
79-88. '
Bamber, D. Reaction times and error rates for Kinchla, R. Detecting target elements in multielement arrays: A
"same"-"different" judgments of multidimensional stimuli. confusability model. Perception & Psychophysics, 1974 15,
Perception & Psychophysics, 1969, 6, 169-174. 149-158. '
Biork, E. L., & Estes, W. K. Letter identification in relation to McIntyre, Coo Fox, Roo & Neale, J. Effects of noise similarity and
linguistic context and masking conditions. Memory & redundancy on the information processed from brief visual
Cognition, 1973, 1,217-223. displays. Perception & Psychophysics, 1970, 7,328-332.
Colegate, R. L., Hoffman, J. E., & Eriksen, C. W. Selective Rumelhart, D. E. A multtcornponent theory of the perception of
encoding from multielement visual displays. Perception & briefly exposed visual displays. Journal of Mathematical
Psychophysics, 1973, 14, 217-224. Psychology, 1970, 7, 191-218.
Eriksen, C. W., & Hoffman, J. E. Temporal and spatial Shiffrin, R., & Gardner, G. T. Visual processing capacity and
characteristics of selective encoding from visual displays. attentional control. Journal of Experimental Psychology,
Perception & Psychophysics, 1972, 12, 201-204. 1972,93,72'82.
Eriksen, C. W., & Hoffman, J. E. The extent of processing of Sternberg, S. Two operations in character recognition: Some
noise elements during selective encoding from visual displays. evidence from reaction time measurements. Perception &
Perception & Psychophysics, 1973, 14,155-160. Psychophysics, 1967, 2, 45·53.
Estes, W. K. Interactions of signal and background variables in
visual processing. Perception & Psychophysics, 1972, 12,
278-286. (Received for publication February 19,1974;
F'lorn , M. C., Weymouth, F. W., & Kahneman, D. Visual accepted March 27, 1974.)