The role of spatial frequency channels in letter identification

Najib J. Majaj, Denis Pelli, Peri Kurshan, Melanie Palomares

Research output: Contribution to journalArticle

Abstract

How we see is today explained by physical optics and retinal transduction, followed by feature detection, in the cortex, by a bank of parallel independent spatial-frequency-selective channels. It is assumed that the observer uses whichever channels are best for the task at hand. Our current results demand a revision of this framework: Observers are not free to choose which channels they use. We used critical-band masking to characterize the channels mediating identification of broadband signals: letters in a wide range of fonts (Sloan, Bookman, Künstler, Yung), alphabets (Roman and Chinese), and sizes (0.1-55°). We also tested sinewave and squarewave gratings. Masking always revealed a single channel, 1.6±0.7 octaves wide, with a center frequency that depends on letter size and alphabet. We define an alphabet's stroke frequency as the average number of lines crossed by a slice through a letter, divided by the letter width. For sharp-edged (i.e. broadband) signals, we find that stroke frequency completely determines channel frequency, independent of alphabet, font, and size. Moreover, even though observers have multiple channels, they always use the same channel for the same signals, even after hundreds of trials, regardless of whether the noise is low-pass, high-pass, or all-pass. This shows that observers identify letters through a single channel that is selected bottom-up, by the signal, not top-down by the observer. We thought shape would be processed similarly at all sizes. Bandlimited signals conform more to this expectation than do broadband signals. Here, we characterize processing by channel frequency. For sinewave gratings, as expected, channel frequency equals sinewave frequency fchannel = f. For bandpass-filtered letters, channel frequency is proportional to center frequency fchannel ∝fcenter (log-log slope 1) when size is varied and the band (c/letter) is fixed, but channel frequency is less than proportional to center frequency fchannel∝fcenter2/3 (log-log slope 2/3) when the band is varied and size is fixed. Finally, our main result, for sharp-edged (i.e. broadband) letters and squarewaves, channel frequency depends solely on stroke frequency, fchannel/10c/deg = fstroke/10c/deg2/3, with a log-log slope of 2/3. Thus, large letters (and coarse squarewaves) are identified by their edges; small letters (and fine squarewaves) are identified by their gross strokes.

Original languageEnglish (US)
Pages (from-to)1165-1184
Number of pages20
JournalVision Research
Volume42
Issue number9
DOIs
StatePublished - 2002

Fingerprint

Stroke
Noise

Keywords

  • Channels
  • Contrast sensitivity function
  • Identification
  • Letters
  • Low-frequency cut
  • Masking
  • Most sensitive channel
  • Noise additivity
  • Object recognition
  • Scale dependence
  • Scale invariance
  • Sinewaves
  • Spatial frequency
  • Spatial vision
  • Squarewaves

ASJC Scopus subject areas

  • Ophthalmology
  • Sensory Systems

Cite this

The role of spatial frequency channels in letter identification. / Majaj, Najib J.; Pelli, Denis; Kurshan, Peri; Palomares, Melanie.

In: Vision Research, Vol. 42, No. 9, 2002, p. 1165-1184.

Research output: Contribution to journalArticle

Majaj, Najib J. ; Pelli, Denis ; Kurshan, Peri ; Palomares, Melanie. / The role of spatial frequency channels in letter identification. In: Vision Research. 2002 ; Vol. 42, No. 9. pp. 1165-1184.
@article{81263d6265754663b15be8cb12fe9e31,
title = "The role of spatial frequency channels in letter identification",
abstract = "How we see is today explained by physical optics and retinal transduction, followed by feature detection, in the cortex, by a bank of parallel independent spatial-frequency-selective channels. It is assumed that the observer uses whichever channels are best for the task at hand. Our current results demand a revision of this framework: Observers are not free to choose which channels they use. We used critical-band masking to characterize the channels mediating identification of broadband signals: letters in a wide range of fonts (Sloan, Bookman, K{\"u}nstler, Yung), alphabets (Roman and Chinese), and sizes (0.1-55°). We also tested sinewave and squarewave gratings. Masking always revealed a single channel, 1.6±0.7 octaves wide, with a center frequency that depends on letter size and alphabet. We define an alphabet's stroke frequency as the average number of lines crossed by a slice through a letter, divided by the letter width. For sharp-edged (i.e. broadband) signals, we find that stroke frequency completely determines channel frequency, independent of alphabet, font, and size. Moreover, even though observers have multiple channels, they always use the same channel for the same signals, even after hundreds of trials, regardless of whether the noise is low-pass, high-pass, or all-pass. This shows that observers identify letters through a single channel that is selected bottom-up, by the signal, not top-down by the observer. We thought shape would be processed similarly at all sizes. Bandlimited signals conform more to this expectation than do broadband signals. Here, we characterize processing by channel frequency. For sinewave gratings, as expected, channel frequency equals sinewave frequency fchannel = f. For bandpass-filtered letters, channel frequency is proportional to center frequency fchannel ∝fcenter (log-log slope 1) when size is varied and the band (c/letter) is fixed, but channel frequency is less than proportional to center frequency fchannel∝fcenter2/3 (log-log slope 2/3) when the band is varied and size is fixed. Finally, our main result, for sharp-edged (i.e. broadband) letters and squarewaves, channel frequency depends solely on stroke frequency, fchannel/10c/deg = fstroke/10c/deg2/3, with a log-log slope of 2/3. Thus, large letters (and coarse squarewaves) are identified by their edges; small letters (and fine squarewaves) are identified by their gross strokes.",
keywords = "Channels, Contrast sensitivity function, Identification, Letters, Low-frequency cut, Masking, Most sensitive channel, Noise additivity, Object recognition, Scale dependence, Scale invariance, Sinewaves, Spatial frequency, Spatial vision, Squarewaves",
author = "Majaj, {Najib J.} and Denis Pelli and Peri Kurshan and Melanie Palomares",
year = "2002",
doi = "10.1016/S0042-6989(02)00045-7",
language = "English (US)",
volume = "42",
pages = "1165--1184",
journal = "Vision Research",
issn = "0042-6989",
publisher = "Elsevier Limited",
number = "9",

}

TY - JOUR

T1 - The role of spatial frequency channels in letter identification

AU - Majaj, Najib J.

AU - Pelli, Denis

AU - Kurshan, Peri

AU - Palomares, Melanie

PY - 2002

Y1 - 2002

N2 - How we see is today explained by physical optics and retinal transduction, followed by feature detection, in the cortex, by a bank of parallel independent spatial-frequency-selective channels. It is assumed that the observer uses whichever channels are best for the task at hand. Our current results demand a revision of this framework: Observers are not free to choose which channels they use. We used critical-band masking to characterize the channels mediating identification of broadband signals: letters in a wide range of fonts (Sloan, Bookman, Künstler, Yung), alphabets (Roman and Chinese), and sizes (0.1-55°). We also tested sinewave and squarewave gratings. Masking always revealed a single channel, 1.6±0.7 octaves wide, with a center frequency that depends on letter size and alphabet. We define an alphabet's stroke frequency as the average number of lines crossed by a slice through a letter, divided by the letter width. For sharp-edged (i.e. broadband) signals, we find that stroke frequency completely determines channel frequency, independent of alphabet, font, and size. Moreover, even though observers have multiple channels, they always use the same channel for the same signals, even after hundreds of trials, regardless of whether the noise is low-pass, high-pass, or all-pass. This shows that observers identify letters through a single channel that is selected bottom-up, by the signal, not top-down by the observer. We thought shape would be processed similarly at all sizes. Bandlimited signals conform more to this expectation than do broadband signals. Here, we characterize processing by channel frequency. For sinewave gratings, as expected, channel frequency equals sinewave frequency fchannel = f. For bandpass-filtered letters, channel frequency is proportional to center frequency fchannel ∝fcenter (log-log slope 1) when size is varied and the band (c/letter) is fixed, but channel frequency is less than proportional to center frequency fchannel∝fcenter2/3 (log-log slope 2/3) when the band is varied and size is fixed. Finally, our main result, for sharp-edged (i.e. broadband) letters and squarewaves, channel frequency depends solely on stroke frequency, fchannel/10c/deg = fstroke/10c/deg2/3, with a log-log slope of 2/3. Thus, large letters (and coarse squarewaves) are identified by their edges; small letters (and fine squarewaves) are identified by their gross strokes.

AB - How we see is today explained by physical optics and retinal transduction, followed by feature detection, in the cortex, by a bank of parallel independent spatial-frequency-selective channels. It is assumed that the observer uses whichever channels are best for the task at hand. Our current results demand a revision of this framework: Observers are not free to choose which channels they use. We used critical-band masking to characterize the channels mediating identification of broadband signals: letters in a wide range of fonts (Sloan, Bookman, Künstler, Yung), alphabets (Roman and Chinese), and sizes (0.1-55°). We also tested sinewave and squarewave gratings. Masking always revealed a single channel, 1.6±0.7 octaves wide, with a center frequency that depends on letter size and alphabet. We define an alphabet's stroke frequency as the average number of lines crossed by a slice through a letter, divided by the letter width. For sharp-edged (i.e. broadband) signals, we find that stroke frequency completely determines channel frequency, independent of alphabet, font, and size. Moreover, even though observers have multiple channels, they always use the same channel for the same signals, even after hundreds of trials, regardless of whether the noise is low-pass, high-pass, or all-pass. This shows that observers identify letters through a single channel that is selected bottom-up, by the signal, not top-down by the observer. We thought shape would be processed similarly at all sizes. Bandlimited signals conform more to this expectation than do broadband signals. Here, we characterize processing by channel frequency. For sinewave gratings, as expected, channel frequency equals sinewave frequency fchannel = f. For bandpass-filtered letters, channel frequency is proportional to center frequency fchannel ∝fcenter (log-log slope 1) when size is varied and the band (c/letter) is fixed, but channel frequency is less than proportional to center frequency fchannel∝fcenter2/3 (log-log slope 2/3) when the band is varied and size is fixed. Finally, our main result, for sharp-edged (i.e. broadband) letters and squarewaves, channel frequency depends solely on stroke frequency, fchannel/10c/deg = fstroke/10c/deg2/3, with a log-log slope of 2/3. Thus, large letters (and coarse squarewaves) are identified by their edges; small letters (and fine squarewaves) are identified by their gross strokes.

KW - Channels

KW - Contrast sensitivity function

KW - Identification

KW - Letters

KW - Low-frequency cut

KW - Masking

KW - Most sensitive channel

KW - Noise additivity

KW - Object recognition

KW - Scale dependence

KW - Scale invariance

KW - Sinewaves

KW - Spatial frequency

KW - Spatial vision

KW - Squarewaves

UR - http://www.scopus.com/inward/record.url?scp=0036235163&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0036235163&partnerID=8YFLogxK

U2 - 10.1016/S0042-6989(02)00045-7

DO - 10.1016/S0042-6989(02)00045-7

M3 - Article

C2 - 11997055

AN - SCOPUS:0036235163

VL - 42

SP - 1165

EP - 1184

JO - Vision Research

JF - Vision Research

SN - 0042-6989

IS - 9

ER -