Eccentric vision : adverse interactions between line segments

(1)

Eccentric vision : adverse interactions between line segments

Citation for published version (APA):

Andriessen, J. J., & Bouma, H. (1976). Eccentric vision : adverse interactions between line segments. Vision

Research, 16(1), 71-78. https://doi.org/10.1016/0042-6989(76)90078-X

DOI:

10.1016/0042-6989(76)90078-X

Document status and date:

Published: 01/01/1976

Document Version:

Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)

Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can be

important differences between the submitted version and the official published version of record. People

interested in the research are advised to contact the author for the final version of the publication, or visit the

DOI to the publisher's website.

• The final author version and the galley proof are versions of the publication after peer review.

• The final published version features the final layout of the paper including the volume, issue and page

numbers.

Link to publication

General rights

Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain

• You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement:

www.tue.nl/taverne Take down policy

If you believe that this document breaches copyright please contact us at: openaccess@tue.nl

providing details and we will investigate your claim.

(2)

ECCENTRIC VISION: ADVERSE INTERACTIONS

BETWEEN LINE SEGMENTS

J. J. .hDRIESES and H. &%%A

lnstituut voor Perceptie Onderzoek. lnsulindelaan 2. Eindhoven. The Netherlands

Abstract-The paper deals with adverse interactions between line stimuli in eccentric vision. Both contrast threshold and just noticeable difference of slant have been measured for a test line as a function of the distance from a number of surrounding lines. Test lines were either parallel or perpendicular to the surrounding lines.

It turns out that the interference affects both contrast threshold and j.n.d. of slant with a clear-cut oBspecificity. The surprising result is the extensive spatial range of the interference: between parallel lines it operates over retinal distances of about O+, degrees, where cp, is the eccentricity of the test line. Large-distance interference limits eccentric spatial &ion in daily life much more than classic visual acuity limits would indicate. and makes eccentric vision probably quite different from “unfocussed” foveal vision.

(1) ISTRODL’CI-IO&

Eccentric or peripheral vision is distinguished from fovea1 vision by a lower spatial resolution. In daylight conditions, the dimension of the resolvable detail of a standard optotype such as a Landolt C increases roughly in proportion to retinal eccentricity, and visual acuity, defined as the reciprocal value of this detail in min of arc visual angle. falls accordingly (Sloan. 1968). In standard acuity measurements the relatively simple test stimulus appears isolated against a homogeneous background. If the test stimulus is complex or if additional stimuli are present. adjacent to the test stimulus, a second difference appears between eccentric vision and fovea1 vision: eccentric vision may suffer considerably whereas fovea1 vision is left largely unaffected. This limitation of eccentric vision was first described for letter strings (Korte.

1923; W~dworth and Schlosberg 1954).

In an earlier experiment. we presented randomly chosen test letters in parafoveal vision and measured recognition scores as a function of retinal eccentricity (a) for isolated test letters and (b) for test letters embedded between two letters /x/ at ordinary type- writer distance. For the embedded test letters correct scores decreased much more with eccentricity than for the isolated test letters. For an equal response criterion, the eccentricity of the embedded letters came out as only one fourth of the eccentricity of the isolated letters. When for embedded letters a larger distance between the three letters was chosen. test scores were higher, but it was not until distances of about half the eccentricity of the test letter were reached, that the scores equalled those of isolated letters (Bouma, 1970). These results suggested to us an explanation in terms of adverse interactions operating over a certain retinal distance rather than in terms of an overloading mechanism (cf. Mackworth. 1965). It has also become clear that, in words, the most outward letter tends to suffer less than the most inward letter. quite in contrast to the common notion

of vision becoming always worse when distance from the fovea increases (Bouma, 1973).

If these effects are due to adverse interactions, it is relevant to ask for any stimulus specificities. since these may indicate at what IeveI of visual signal processing the interaction occurs. In our first esperiments with letters. we observed that the letter ,vi suffered more than certain other letters when embedded between two letters /xi, as is demonstrated in Fig.

I. This suggests that it may be the parallel oblique line segments which interfere mutually. Such an orientational specificity would fit in nicely with neurophy- sioiogical views (Hubel and Wiesel. 1965. 1968). This led us to experiments in which the orientation of a line segment had to be perceived in the presence of other line segments. In eccentric vision. the expected orienrational specificity of the interference was indeed observed (Beerens and Bouma, 197%). We continued with experiments on perceptual limits of slant perception in an interference situation. Consequently, we report here on the influence of surrounding line segments on (a) the just noticeable difference &n.d.) in orientation

of

a test line and (b) the contrast detection threshold of a test line, both as a function of the distance from interfering line segments, either parallel to the test line or perpendicular to it. Some additional observations concern the influence of the retinal

V . xvx xvx * V

a . xax

xax . a

Fig. I. Adverse interaction in eccentric vision. In looking successively at the dots, the recognizability of the, embedded letters is hampered by the flanking ‘xi letters. Letter

(3)

12 J. J. ANDRESSEN and H. BOUM eccentricity of the test line (Andriessen. Eburna and

Ekerens. 197 1).

(2) JUST XOTICEABLE DIFFERENCE OF SLAM-

Apparatus

At a distance of 57 cm from the observer’s eyes a fixation dot (F) and a test line surrounded by parallel lines were projected on to a circular and translucent verticai screen (Fig. 2).

The observer used his right eye only with natural pupil and focussed on the fixation dot. The head was held upright. being steadied by a rest consisting of a chin sup- port, two temple steadiers and a forehead rest. Conse- quently. the presentation of the test line and the surrounding lines was at the temporal side of the retina on and around the horizontal meridian. The eccentricity was constant, usually at cp, = 12”. The slant z of the test line was variable. The primary slant was x = 135” and in order to find a just noticeable difference in slant, the slant was made adjustable to either side. Slants of the surrounding lines were either /) = 135’ (parallel situation. 11) or b = 4.5” (perpendicular situation. I). Test line and surrounding lines were projected as luminous line segments (AL. = 8 cd/m’), against a background of L, = 3 cd/mz. The total luminance L, of the test line was thus I I cd/m’. The ring of surround- ing lines was projected by means of a slide. one for each value of p.

The test line was projected as follows: A slit, situated centrally in a rotatable disc, was projected onto the translucent screen. Rotation of the disc was achieved by a step- motor device in steps of about 0.1’. Mechanical clearance in the gear-box was negligibly small. Both the slide projec- tor and the projecting device for the test line were provided with electrically controlled shutters. which were triggered synchronously.

The surrounding lines subtended 1.2’ x O-2’ and the test line subtended 05” x 0.1”. Pilot experiments indicated that at these dimensions adverse interactions were most expressed. Remarkably. longer surrounding lines produced less interaction.

perpendicular

I

c

PLi

4

Fig. 2. Stimulus configuration: Fixation dot F, test line (5 x 1 mm) and surrounding lines (12 x 2 mm). cp,--eccentricity of the test line; distance between test line and surrounding lines (spacing): z-slant of the test line, z = 135’; /3--slant of the surrounding lines, either b = 135”

or /I = 45”.

’ Experiments with x - AZ slants give basically the same results. I0

dsg

15 +

Y2

++

T

9 1 -+ + 6 =+ - + -at 3 - - 01 1 3 5 7 9 11 number of series 2 4 6 0 10 12

Fig. 3. Example of a score diagram. Each datum is based on I2 stimulus pairs at most. For details. see text. Procrdurr

By depressing a push-button, the observer started the time cycle in which the test line and the surrounding lines were presented twice. each time for lo0 msec and with an interval of I.5 sec. A pilot experiment showed that this time interval was not critical. The first time the test line was always presented with the primary slant of x = 135’. The second time the slant was randomly either I = 135’ again or x + AZ.’ AZ being variable between about 0.1’ and 33’. The task of the observer was to indicate whether or not he noticed a slant difference (rotation) between the first and the second presentation. The subject responded corres- pondingly by depressing one of two push-buttons (forced choice).

The value of AX depended on the scores. according to a sequential up-and-down method (Cardozo. 1966). The value of Ax was increased when too many errors occurred; AZ was decreased when the response criterion was exceeded. We aimed at a correct response criterion of 75% as the threshold. The sequential up-and-down method is designed so as to present most pairs of stimuli just above and below this threshold. Use was made of apparatus described in Cardozo and De Jong (1968). The method is characterized by short series lengths of I2 pairs of stimuli at most. The procedure is as follows:

As soon as in a series the correct response criterion of 75% is signifmantly exceeded, the series is automatically interrupted. A positive score (+ in Fig. 3) is counted if five out of five. seven out of eight or nine out of II responses are correct. Ax is then decreased by one step of 3”. A negative score (-) is counted if more than three errors occur in a series. In this case Ax is increased by 3”. If precisely three errors out of I2 responses are counted (= ). the series is repeated with the same Ax. Figure 3 gives an example of such a series. Series are continued until at least three transitions from + to - have occurred. An estimation of the threshold value of Ax is taken to be the mid point of the transition from + to -. Ax, is calculated as the average of the consecutive mid point positions. We define A.xx, as the just noticeable difference of slant; abbre- viated as j.n.d. of slant. (To be exact, this threshold corresponds with a correct response criterion of 7lo/ Cardozo, 1966.)

During one experimental session one value of Ax, was determined for each of the following p-values (see Fig. 2): 2”. 3’. 1”. 5’. 6’. 7’. 9’ and r (no surrounding lines). The sequence of p was random.

In a number of training sessions the observers got used to the set-up and learned to concentrate on the eccentric position of the stimulus array. while maintaining fixation. The importance of this locus of attention has recently been shown by Engel (1971). Afterwards, at least four exper-

(4)

Eccentric vision

-P -P

Fig. 4. Upper graphs: J.n.d. of slant AT, of the test line as a function of distance 0 from surrounding lines. either parallel (11) to the test line or perpendicular (I) to the test line. Two observers. Test eccentricity 9, = I, 7’ . Lower graphs: The difference A?,, - &, as a function of 1’. Vertical bars indicate

95% confidence intervals.

13

imental sessions were carried out for each situation, alter- nately /) and 1 sessions.

Two observers (JvdL and JvdP), with normal vision, took part in the present experiments. Pilot experiments, which gave similar results were carried out by five other observers.

Results

The results are summarized in Fig. 4. In the upper graphs the average value of AZ, is plotted as a function of the distance p between test line and surrounding lines for the situations I/ and 1.

It appears that the surrounding lines increase the j.n.d. of slant for p 6 5” for both observers. For the parallel situation the effect is large, but also for the perpendicular situation and small values of p, j.n.d. of slant is increased. The value of AZ, for p = 5” (observer JvdP) is slightly lower than at p = X. We have noticed similar small effects in a few other observers, though the differences were not significant. In the lower graphs of Fig. 4, the differences between the data points of the top curves, denoted by A% - AZ1 have been plotted, again as a function of p. These curves show how j.n.d. of slant of a test line increases when the surrounding lines change over from a perpendicular to a parallel position with respect to the primary test line.

(3) CO!VTRAST THRESHOLD

Apparatus

The apparatus was practically the same as in the above experiments. There are two differences: (a) the slant z of the test line is now constant at z = 135’. (b) at the same background luminance L,, of 3cd/m’. the luminance AL of the test line itself is now the independent variable. It can be varied between @25 and 93 cd/m’ by means of neu- tral density filters in steps of about 200/,. The total

luminance of the test line is accordingly from 3.25cd/m’ upwards. Test line contrast C was defined as A&‘&,. Procedure

We chose the experimental procedure similar to the procedure of the experiments on j.n.d. of slant. The two IO0 msec stimuli (a) and (b) were now successively: (a) surrounding lines together with test line. and (b) surrounding lines, randomly either with or without test line. The observer indicated by pressing push-buttons whether he had noticed the second presentation of the test line.

Again we used the sequential up-and-down method. From the results the 75% contrast threshold C, was calculated. For each of the experimental conditions jj and 1. four sessions were carried out in the same way as described in Section (2). The same two observers took part in the experiments. In pilot experiments two other observers took part.

Results

Figure 5 shows the average value cz of the contrast threshold of the test line as a function of the distance p between test line and surrounding lines for the two experimental conditions /) and 1.

Again the influence of parallel surrounding lines is clearly noticeable for p 5 5’. For perpendicular surrounding lines the influence on c, is small, except at p = 2’ for observer JvdL.

(4) CONTRAST-COMPENSATED J.N.D. OF SLANT

The influences of surrounding lines on the j.n.d. of slant of a test line (Fig. 4) and on the contrast threshold of a test line (Fig. 5) look rather similar. This raises the question whether the two phenomena are directly related in the sense that the increase of j.n.d. of slant might perhaps be caused by an increased contrast threshold. This led to a new series of experiments on the j.n.d. of slant in which the

(5)

‘4 _{J J.}.~SDRiESSES and H. Boc->U 0.5 - J.v.d. I. we ohox 0.4

Ol

2 5 IO A v, deg. -? 0.5 - J.v.d.f? 0.4 I/ -0.3. \

Fig. 5. Contrast threshold cc of the test line as a function of distance p between test line and surrounding lines for the experimental conditions I) and i. Two observers. Test eccentricity cpc = II’. Luminance of the background of the test line L, = 3cd,‘m’. The total luminance of the sur-

rounding lines was constant at 11 cd, m’. luminance AL = 8 cd/m’ of the test line was increased to a value A’L in order to compensate for the increase of the test line contrast threshold. .4t each distance p between test line and surrounding lines.

M 15 I$ lo I 5 0 J.v.d.L.

as prescribed by the individual T, curves of Fig. j.

Results

The upper graphs of Fig. 6 show the result of the experiments. Again. AZ, has been plotted as a function of p for the parallel and ~~endicu~ar slants of the surrounding lines. The lower graphs show the differences A?, - A?_ as a function of p with 95’” confidence intervals. The dashed lines indicate the lower graphs of Fig. 4 at AL = Ycd m2. It appears that there is some decrease in jnd. of slant at p 5 5’, but it definitely does not reach the values of AYc at p = X. There remains then a definite. orien~tion specific. adverse influence of the surrounding lines on the j.n.d. of slant. larger for parallel lines than for perpendicular ones,

15) .+DDITIOSti ESP&RI>IEXTS

The above experiments were all done at one eccentricity qr = 12’ and at one primary test line slant at r = 135’. The surrounding lines were either /3 = IX’ (parallel condition) or /3 = 45’ (perpendicular condition). In additional j.n,d. experiments these parameters were varied. For reasons of etIiciency. we re- stricted ourselves to small values of y and the parallel condition. Otherwise, the experiments were quite similar to those of Section (2) (no contrast compensa- tion). Four observers, different from those in the other experiments took part in these experiments. Also, for two observers we rotated the whole stimulus configur-

20 15 1% 10 i 5 0 ~ J.ud.f! -P

Fig. 6. Contrast-compensated experiments. tipper graphs: J.n.d. of slant A?, of the test line as a function of distance p for the experimental conditions /j and i. Two observers. The contrast of the test line was increased in order to compensate for the increase of the contrast threshold (see tent). Test eccentricitv qK = 12’. Lower graphs: The difference Ai - Ari_ as a function of p. Dashed iines refer to the

(6)

Eccentric vision 0’ ₁ ₃ ₁₀ _Jo’..A d?g. -P 30 . IQ qap3.5’ I 20 ’ ‘R-6. qlpw 0 _{_____} 10. -4% \---+-a \ ‘, >______J ____ .+_a O- 1 3 10 30-k deg. 40 deg. -P 30 I deg. 0.1 a3 1.0 3.0 k - P/q

Fig. 7. Upper graphs: J.n.d. of slant A?, of the test line as a function of distance ,D. TWO observers. Parallel condition only. Test eccentricity cpI was 6”. 12’ and 36’. Stimulus dimensions were chosen proportional to qI. Lower graphs: Values of AT: (corrected for vertical shift at p = x. arbitrarily towards the value for cp, = 12”) as a function of the relative distance p/cp, between test line and surrounding lines. At qt = 36’ and p = 6’ (p/q, = 0.17) Ax had to be in excess of Ar mill = 33’ for observer

JA.

ation in order to see if horizontal and vertical orientations gave results similar to oblique orientations.

Eccwtricify cp

The experiments were carried out at eccentricities rp, = 6’. 12” and 36’. and with parallel surrounding lines only. In order to account for the decrease in visual resolution with increasing eccentricity the visual angles subtended by the stimulus configuration (Fig. 2) were chosen proportional with p,. Below e+ = 6”, however, the visibility of the test line became too poor. With cp, = 12” and cp, = 36’ the viewing distance was 57 cm. with p, = 6’. however. the viewing distance was I I4 cm.

The experimental results of two observers are summarized in Fig. 7 (upper graphs). The values of AZ-, for isolated test lines (p = r) turn out slightly different. Since we are interested in the influence of p. we have corrected for these vertical differences by a vertical shift. thus plotting

A$. = Ai, - [Ax, - Aa,@= ,,.I,= z. If we plot horizontally values of p/cp, (Fig. 7, lower graphs) it turns out that at p/cp, 2 04-O-7 the influence of the parallel surrounding lines is negligible, No general conclusion can be drawn, however, about the relationship between the j.n.d. of slant AZ, and p/p,. Two out of four observers give results of the type of Fig. 7 (left) where a general relationship is indicated between j.n.d. of slant and p/p,. However, the two other observers give results of the type of Fig. 7 (right) in which the curves branch off. Only the first type admits a simple interpretation: the criti-

cal distance on the retina between interacting stimuli, increases linearly with eccentricity.

In these experiments the slant of the test line was 2 = 0 (horizontal H). x = 135’. 2 = 90’ (vertical V). In all situations, the surrounding lines were parallel to the test line, hence /3 = 0’. 135” and 90’, respect- ively. The eccentricity cpI was 12’.

J.A.

Fig. 8. J.n.d. of slant A?, of the test line as a function of p. for three main orientations of the whole stimulus configuration: horizontal H. vertical V and oblique. Paral- lel condition. Two observers. Test eccentricity cpI = 12”.

(7)

Figure Y gives the experimental results. The main conclusion is that the adverse influence of surrounding lines on the j.n.d. of slant of a test line is basically ths same for horizontal. oblique and vertical slants of the stimulus configuration. Looking in morz detail. it appears that for observer JA the H slant gives sig- nificantly lower values of the j.n.d. of slant at small values of p than the oblique slant.

kve were a little surprised to find the j.n.d. of slant for oblique orientations to be similar to those for W and v. In fovea1 vision. the j.n.d. of obliques are about twice as high as those of H and L’.

(6) DISCL SSION

In eccentric vision the perception of a test line is hampered by line segments in adjacent rctinnl areas. Interference IS most if the adjacent lines are parallel to the test line. but also if they are perpendicular to it. there is some interference (Fig. -I). Thus. the interference is orientation-specific. The interference between parallel lines decreases with increasing spacing. However. its spatial extent is impressive. since it remains present up till spacings p of approx 040, (Fig. 7). if qr is the eccentricity of the test line. in the following we shall discuss successively (a) the criterion that the observers use in the j.n.d. of slant experiments. (b) the relationship between the increase of the contrast threshold and of the j.n.d. of stant. (c) the orientational specificity of the interference and (d) finally. as to the wide spatial extent of the interference. we ask for related phenome~ in the literature. (a) J.~.rl. crirrrior~

In the j.n.d. of slant experiments three strategies seem to be available. (1) According to the definition 0fj.n.d. the observer is supposed to notice a difference in slant of the test line at the first and at the second presentation. However. the observer can identify the second of the two successive test lines as being less steep than 45’. he does not need the first test line at all. In this case (2) he can use an absolute internal criterion. or (3) he compares the slant of the test line with the slant of the interfering lines. According to the subjective impression of thz observers. the first strategy is not always used. in particular if the second test line differs considerably from 45’. Results based upon the strategies (2) and (3) may be considered as just noticeable differences all the same. It will be evi- dent that judgments of slant differences require perception of two slants. whereas judgments of absolute slant require the perception of only one slant. In case of judgments of slant differences, the interpretation will have to account for an uncertainty in perceived slant. In case of absolute judgments. the measurements would rather indicate the limit of slant which just escapes from being masked by the interfering lines.

I W’c have abstained from experiments with smaller valuss of p due to the present dimensions of test line and surrounding lines. Also at small values of p the problem of stray light arises. With the present p-values. stray tight from the surrounding lines has little intluencc on the visibility of the test line.

The skperiments indicate that the interference has at least two el%ects on the test line: (I) the contrast threshold is incrsased (2) the j.n.d. of slant is increased. Can a single mechanism be responsible for both phenorneila..’ The similar action range suggests this to be the case. However. when we corrected for the increased contrast threshold by a corresponding increase of luminance. the increax of j.n.d. of slant was not completely compensated {Fig. 6). The hypothesis of a single mechanism can be maintained if it is assumed that luminance does not play a significant role at the: level of visual processing where the inter- I’crencc occurs.

Slant-specific interference between line segments is nou ~vell known in perception research. Seurophysio- logica! evidence strongly suggests that orientation- specific effects are of cortical rather than of retinal origin (Hubel and Wiesel. 1965. 1965). In line with this. the present interference occurred equally in dichoptic pilot experiments. in which the test line was presented to one eye and interfering lines to the other (Beerens and Bourna, 19%).

In Fig. 4 the angular range of the interference extends to & = 18’ at p = 2’ and there are no signs that at smaller distance than p = 2’ the function flattens.’ Some interference is indicated even for per- pendi~~llar test lines. Campbell and Kulikowski (1966). in experiments on orientation-specific masking also found rather large interaction angles.

A f<~ qears ago. slant dependencies of effects such as thzse bverc interpreted straightforwardly in terms of tuning characteristics of human orientation sensors on i\ hich there is an extensive literature now. The j.n.d. of slant would then be dependent on the sharp- ness and the relative amplitudes of the tuning curves involved (Andrews. 1967; Bouma and Andriessen, 196Sa. b). Any single retinal stimulus, however. will excite a great many sensors. possibly of different tapes. and accordingly one can hardly expect such simple relationships between tuning characteristics on the one hand and orientation dependent maskings such as mentioned above on the other. Present theoretical positions are clearl: more reserved and more complex relationships with more parameters are being considered. such as suppression effects (“inhibition”) benveen sensors of differing tuning slant (Andrews. 1967; Blakemore. Carpenter and George- son. 1970: Kufikowski and King-Smith. 1973), or a combination of adaptation and facilitation (Bouma and -\ndriessen. 1970). In neurophysiology there is some initial evidence on orientation-s~cific inhibition processes (Blakemore and Tobin. 1972; Bene- vento. Creutzfeldt and Kuhnt 1972).

As to the present probIem it would be relevant to know if orientational specificity and interaction bshvren hypothetical sensors would depend on retinal eccentricit>. Esperimentally. we have found in pilot experiments that j.n.d. of slant of isolated line segments depend little on retinal eccentricity (Andriessen and Bouma. 1970). However. Sharpe and Tolhurst (1973a. b) have reported for stationary gratings that adaptation effects extend to considerably larger angles ;It an eccentricity of IO- (1973a) than at the fovea

(8)

Eccentric vision 77 (1973b). It would certainly be of interest to determine

if perhaps the coarseness of orientational vision shows less increase with eccentricity than the coarseness of spatial vision.

(d) Spatial esmr of rhr irmrfirrwc

The intriguing aspect of the interference is its wide spatial extent. In terms of the test line at ‘P,. parallel lines hamper its proper perception at distances up to 04 cp,.

This is a wide range of inHuence when compared with ordinary spatial resolving power for isolated objects, (visual acuity)- ‘. which is about 0.01~ (le Grand. 1967; Sloan. 1965). Would all surrounding lines contribute equally to the interference? We have some evidence that outward lines. eccentric to the test line contribute more than inward lines. at the fovea1 side of the test line (Beerens and Bouma. 1970b). This inward-outward asymmetry would suggest that the action range of the interference should perhaps be expressed in terms of the most outward surrounding lines at ((P~),,,~% (see Fig. 2). We then arrive at a spatial action range of about 03((pi),,,.

The present results on line stimuli fit in with esperiments on eccentric letter recognition. mentioned earlier (Bouma, 1970). in which interference was found up to distances of about 0.5 cp? In experiments on recognition of letters from unpronounceable letter strings. we also found an inward-outward asymmetry: inward letters close to the fovea gave lower recognition scores than outward letters farther from the fovea (Bouma. 1973). In these letter experiments, we found in addition. lower recognition scores in the left visual field than in the right one. For the present line stimuli we have no expenmsntal evidence as yet on such a possible left-right difference.

The interference between line elements. in particular if they are of similar slant. may well constitute a major component of the interference between adjacent letters. Engel (1974) has found a luminance- specific interference and a size-specific interference in eccentric vision, both probably also operating over large interaction distances. The implications are that in relatively complex stimulus configurations. the vision of details decreases much more sharply with eccentricitv than classic visual acuity with isolated optotypes indicates.

On large-distance interaction there seems to be only a limited perceptual literature. Atkinson (1973) found for the after-images of two bars at an eccentricity of 2’ an increase of synchronous visibility for distances up to 2’ between parallel bars, and up to I’ for perpendicular bars. Kulikowski and King- Smith (1973) conclude from experiments on subthreshold summation to a fovea1 grating detector ex- tending over about I’. and to a “line detector” with a width of about @I 5’, as compared to fovea1 detail vision of about O-01 ‘. This fovea1 ratio of over 1 j:l can perhaps be compared with the present ratio of 40: I in eccentric vision. though the comparison leaves open the question why interference in fovea1 vision is so much less than in eccentric vision.

Neurophysiological research offers more points of contact. Sizes of receptive fields of orientation-specific cortical units are much larger than corresponds with the supposed local acuity limits and inhibiting areas

(Ranks) may extend to considerable distances from the receptive field center. For the cat. Hubel and Wie- se1 (I 965) report the average area A of receptive fields of comples cells roughly proportional to eccentricity: A (sq. degr.) (:)+o (degr.). For the monkey a similar field organization has been found, but receptive field dimensions are smaller: for the parafoveal region between (3 = I3 and 9 = 1’ linear dimensions of complex receptive fields are estimated as roughly 4 of those of cats (Hubel and Wiesel, 1965). This value for monkeys is smaller than would fit in with the present psychophysical results.

Thinking in terms of complex cells. two possible mechanisms of interference come to mind. (I) Com- plex cells are knolvn to react to one stimulus line almost the same as they do to two parallel stimulus lines if both are within the receptive field. This might be called “saturation’*. In terms of our experiment the interfering lines might have produced so much activit!. that the test line could not properly be detected. (2) The complex cells may show inhibitory flanks esttnding over large distances-in our experiment this would correspond to a partial suppression of activity brought about by the test line. The obser- vation that the interference may be counteracted to a certain extent by increased contrast of the test line. can better be reconciled with a suppression than with a saturation hypothesis.

Finally we would like to draw attention to the many peculiar aspects of eccentric vision. Even relatively simple stimuli may look quite different from what they look like in fovea1 vision (for certain lower case letters. see Bouma. 1971). Perceived localization of stimulus parts may be different from the actual stimulus configuration. The colour of a coloured object may be perceived as belonging to an adjacent object (Bouma. 1969). Eccentric vision. therefore, is much diflerent from a coarse or “unfocussed” type of fovea1 vision. Eccentric vision is an indispensable component of daily vision both for recognizing and for selecting visual objects. We hope that this paper may contribute to a greater research interest in eccentric vision.

clck1ro~~lr[5ern~11rs-~~e gratefully acknowledge the dedi- cated work of our observers: Messrs. E. G. J. Beerens. J. v.d. Leeuw. J. v.d. Pant. H. R. Adema. J. Vissers (all students at Eindhoven University of Technolog?. the Neth- erlands) and .A. L. M. van Rens. Mr. E. G. J. Beerens took an active part in the pilot experiments. Mr. F. L. Engel and Dr. C. A. A. J. Greebe helped us by commenting on an earlier draft.

REFERESCES

Andrews D. P. (1967) Perception of contour orientation in the central fovea-l: Short lines. Vision Rrs. 7. 97s 997.

Andriessen J. J. and Bouma H. (1970) Just noticeable differences in slant of test lines as a function of retinal eccentricity. I.P.O. .-lnmtal Progress Report 5. I IQ-1 14. Andriessen J. J.. Bouma H. and Beerens E. G. J. (1971)

Interaction between lines in eccentric vision: contrast threshold and just noticeable difference in slant. I.P.O.

.III~IIUI~ P~o~w\.T Report 6. 56-60.

Atkinson J. (1973) Properties of human visual orientation detectors: a new approach using patterned afterimages.

(9)

J. J. ASDRESSES and H. BOCW~

Beerens E. G. J. and Bouma H. (19%) Orientational specificity of line interaction in eccentric vision. I.P.O. .-lmuul Proywss Rrporr 5. I I C IN.

Becrrns E. G. J. and Bourna H. (iY7Ob) Asymmetry and symmetry in eccentric line interaction. I.P.O. .-lnnual Pro- gress Report 5. ID- 1’1.

Benevento L. A.. Creutzfeldt 0. D. and Kuhnt U. (1972) Significance of intercortical inhibition in the visual cortex. .Varrrrr. .Vrw Biol. 238. 12-L 116.

Blakemore C.. Carpenter R. H. S. and Georgeson M. .A. (1970) Lateral inhibition between orientation detectors in the human visual system. .~nrurr. Lord. 228. 37-39. Blakemore C. and Tobin E. .A. (lY72) Lateral inhibition

between orientation detectors in the cat’s visual cortex. Esp[ Btuin RCS. 15. J39-UO.

Bouma H. (1969) Visual isolation in eccentric form vision: the role of colour. I.P.O. .4~1rr~rl Progrrss Rrporr 1. 9C 99.

Bouma H. and Andriessen J. J. (196Ya) Perceived orientatron of isolated line segments. C’isio~ Rrs. 8. 195- 507.

Bouma H. (1970) Interaction cfects in parafoveal letter recognition. .Vcrrurc. Land. 226. 177-l 78.

Bouma H. (I Y7l) Visual recognition of isolated lower-case letters. lision RCS. II. 45~47-I.

Bouma H. (1973) Visual interference in the parafovral recognition of initial and tinal letters of uords. C’isiou Res. 13. 767-782.

Bouma H. and Andrirssen J. J. (196Sb) Perceived orientation of line segments. A theoretical contribution. I.P.O. -LII~I~ Prog~ss Rrporr 3. SI-85.

Bouma H. and Andriessrn J. 1. (1970) Induced changes In the perceived orientation of line segments. I’ision Rrs. 10. x-319.

Campbell F. IV’. and Kulikowski J. J. (1966) Orientational selectivity of the human visual system. J. Plt~siol.. Loud. 187. 437-US

Cardoro B. L. (1966) A sequsntral up-and-down method. I.P.O. Ammal Proyrrss R<port I. I I&l 14.

Cardozo B. L. and de Jong Th..-\. (lY6Y) .?I digital ample- menration of the sequential up-and-do&n method. I.P.O. .A~:i?:d PI-oyw.ss Rupovr 3. I I 5-I I 7.

Engrl F. L. (I9711 Visual conspicuity. directed attention and retinal locus. t isiou Rrs. II. 563ei76.

Engsl F. L. (197-t) Visual conspicuitv and selective background inter&exe in eccentric v&ion. I,‘isiou Rrs. 14. 15~47 I,

HUM D. H. and VVirsel T. N. (lY65) Receptive fields and functional architscture in two nonstriate visual areas (13

and 19) of the cat. J. .Sruropir~siol. 28. X9-XY. Hubcl D. H. and VViesel T. N (1968) Receptive fields and

functional architecture of monkey striate cortex. J. P/I)- sioi. LOlfd. 19s. 3 112-G.

Sharp< C. R. and Tolhurst D. J. (IY73aJ Orientation and spanal frequency channels in peripheral vision. I rsioo

R,s 13. ?I().~~71 I?.

Korte U’. (1922) Die Gestaltauffassung im indirecten Schcn. 2. P.SJC/IO(. 93. IT-S2.

Kultkowskt J. J. and King-Smith P. E. (1973) Spatial ar- rangsmcnt of line, edge and grating detectors revealed by subthreshold summation. L’ision Rrs. 13. I-tSill7Y. Ls Grand Y. ( 1967) For/n rr~!rl Spacr I ision pp. 1% 1-W

Indiana Cniv. Press, Bloomington.

Mackuorth N. H. (IY6j) V’isual noise causes tunnel vision. Pc~iilorlo,ri. sci. 3, 67-68.

Sharpe C. R. and Tolhurst D. J. (1973b) The effects of temporal modulation on the orientation channels oi the human vrsual system, Ptwvpriotl 2. 72-Y.

Sloan L. L. (196s) The photopic acuity-luminance functton vvrth special reference to parafovsal vision. t’irioft Rrr. 8. 9~)l~Yl I.

VVooduorth R. S. and Schlosberg H. (1931 Expcvimmd P~~~ci~o~og~~. Methuen. London.