Corrections for
Vision Science: Photons to Phenomenology
Stephen E. Palmer

        The following errors have been identified in the initial printing. Please correct your copy as indicated below. Text in bold indicates the word or phrase that is to be corrected or inserted.
If you find any additional errors, please report them to the author by email at or by mail at the Department of Psychology, University of California, Berkeley, CA 94720-1650.

Title page:
"Stephen E. Palmer
Department of Psychology and
Institute of Cognitive Studies
University of California, Berkeley"

 p. xix, column 1, line 39 should be:
"... that might well have been omitted by nonpsychologists ..."

p. 101, column 2, line 38 should be:
"... complicated and is described in Appendix C, which explains it..."

p. 104, column, line 20 should be:
"... numbers 3 and 42 in the dot patterns..."

p. 117, The gray regions in Figure 3.2.19A are too dark. It should look like this:



p. 124, in Figure 3.3.2: the formula at the top incorrectly gives the symbol for Illumination Spectrum as R(sub)w, but it should be I(sub)w.

p. 159, Figure 4.2.1 caption should be:
"... (B) a higher spatial frequency, ..."

p. 166, lines 9-12 should be:
 "... so that the low-frequency grating always stimulates the upper half of your retina and the high-frequency grating
     always stimulates the lower half."

 p. 167, column 2 to p. 168, column 1:
The experiment described in detail here and incorrectly attributed to Graham and Nachmias (1971) was actually performed by Campbell and Robson (1968). (See p. 740 for the complete reference.)

 p. 173, column 1, line 32 should include a footnote:
"... convolution of an edge operator with an image.* ..."

* Technically , the operation described on pages 173 and 174 is the cross correlation of the edge operator and the image, but it is usually referred to as convolution in the computer vision literature (see Nalwa, 1993, p. 66). The convolution is actually the cross correlation of the image with the reflection of the edge operator about its center. Cross-correlation and convolution are therefore identical operations for any symmetric edge operators (e.g., Figures 4.3.1 C, D, and E), but not for asymmetric operators (e.g., Figures 4.3.1 A and B). Convolution is preferred in the computer vision community because of several important mathematical properties, such as commutativity, associativity, and distributivity over addition.

p. 215, Figure 5.3.12 caption should be:
"... in the text, a checkerboard of squares can be seen floating ..."

 p. 221, Figure 5.3.19 caption should be:
"... The images on the far right were reconstructed from the responses of multiorientation, multiscale spatial filters positioned at the center of the images on the far left. ..."

p. 221, column 2 line 8 should be:
"... Optic expansion results, for example, when you walk ..."

 p. 263, column 1, line 16 should be:
"... For example, Palmer and Back used the repetition discrimination task ..."

 p. 285, column 1, line 4 should be:
"... because object recognition requires a candidate object (namely, the ..."

p. 286, column 1, line 14 should be:
"(See Casati & Varzi, 1994, for a discussion..."

 p. 292, Figure 6.4.7 caption should be:
"... A white rectangle occluding four octagons is perceived in part A, ..."

 p. 308, Figure 6.6.4 caption should be:
"... Five-month old infants fail to show dishabituation ..."

 p. 315, column 2, Equation 7.1 (and in Figure 7.1.1):
Note that this equation is not exactly the same as the equation on page 233 and in Figure 5.5.9 because the line of sight goes to the center (rather than the bottom) of the object.

 p. 339, footnote 3 should read:
"Because the image on the retina is fully inverted (including left-right reversal), ..."

p. 375, footnote 2 should read:
"... then the longest dimension and center of mass can change substantially ..."

p. 379, column 1, lines 17-18 should read:
"... yet strongly influence the correlation between a template and an input image."

p. 435, Figure 9.3.2 caption should read:
"... and aspect ratio (H and I). (R&R = rotational and reflectional symmetry.)"

p. 485, column 1, line 5 should include a footnote:
"... a version of which is shown in Figure 10.1.18 A*.
    * Barlow and Levick (1965) also discussed a version that uses inhibitory rather than excitatory connections to
    produce directional selectivity. Although it is perhaps less intuitively clear, it provided a better fit to their data
    The interested reader is referred to their original article for further information.

p. 491, column 1, line 15 should read:
"... by photocopying the page facing the inside back cover ..."

 p. 534, footnote 2 should read:
"... rather than that from the attended channel..."

 p. 556, column 2, last line should read:
"... (e.g., disconnected redness and horizontalness); we perceive ..."

 p. 587, col. 2, 4th line from the end should read:
"... such as inattentional blindness, the attentional blink, and change blindness (see Section 11.2.1).

p. 635, Figure 13.2.4 caption should read:
"... seeing). (Data replotted from Weiskrantz et al., 1974.)"

 Color Plate C.2
Technical problems resulted in this figure being misprinted with thin white lines between the blue and yellow squares,
which destroy the intended illusion. The correct figure can be viewed above .
Viewers are forewarned that differences between different color monitors may weaken the illusion.