A new ‘CFS tracking’ paradigm reveals uniform suppression depth regardless of target complexity or salience

  1. David Alais
  2. Jacob Coorey
  3. Randolph Blake
  4. Matthew J Davidson  Is a corresponding author
  1. School of Psychology, The University of Sydney, Australia
  2. Department of Psychology, Vanderbilt University, United States
6 figures, 1 table and 1 additional file

Figures

The critical measure of suppression depth remains untested in CFS.

(a, b) Exampe stimulus displays for the discrete trials, during which (a) a low-contrast target steadily increases in contrast until visibility is reported (bCFS), and (b) discrete reCFS trials, with a high-contrast target decreasing until target invisibility is reported. (c) A typical bCFS result, in which a target image (here a face or non face) is initially weak and steadily increases until it breaks suppression. Faster breakthrough times (lower contrast) for face stimuli are often interpreted as evidence that faces undergo expedient processing that counteracts their susceptibility to suppression relative to other visual stimuli. Without also measuring suppression thresholds for each stimulus, this conclusion is premature. (d-f) To define the magnitude of suppression during CFS, it is necessary to measure the contrast thresholds at which stimuli enter and exit awareness, with the difference indicating suppression depth. Red bars display hypothetical results measuring the contrast at which an initially visible stimulus with decreasing contrast becomes suppressed by the mask (reCFS). The results of panel c are reproduced in d-f in light blue. (d) If the reCFS thresholds for face and non-face images are the same, this would support reduced suppression depth for faces (as δface < δ non-face). (e) Alternatively, the reCFS thresholds for faces and non-faces might differ, with the face remaining visible at a lower contrast than non-face images (lower reCFS threshold), indicating more suppression for faces than non-faces (as δface > δnon-face). (f) Finally, the reCFS thresholds might differ between faces and non-faces but by an amount equivalent to their bCFS differences, indicating the same suppression depth for both image types (as δface = δnon-face). Such a result would argue against enhanced unconscious processing of face stimuli.

Example tCFS trial and comparison to discrete conditions.

(a–b) Example tCFS trial from one participant, showing the change in contrast over time. Red markers indicate the level at which a target with decreasing contrast became suppressed (reCFS) and blue markers indicate traditional bCFS responses (breakthrough of contrast-increasing target). The same trial is shown with target contrast in decibel scale in b. Because the human visual system has a logarithmic contrast response, it is appropriate to increase/decrease target contrast logarithmically as in b, to create a contrast change that is perceptually linear. (c) Interaction between thresholds and condition type for the 20 participants. Individual blue and red dots display participant means for bCFS and reCFS respectively. Grey lines link thresholds per participant, per condition. Blue and red diamonds display the mean across participants, and error bars plot ± 1SEM corrected for within participant comparisons (Cousineau, 2005).

Suppression depth is uniform across image categories.

(a) bCFS and reCFS thresholds by image type for the 18 participants. Blue and red diamonds display the mean across participants, and error bars plot ±1 SEM corrected for within participant comparisons (Cousineau, 2005). Grey lines link thresholds per participant, per image type. For visualization, we have linked the bCFS and reCFS thresholds with broken lines, to better indicate that both do vary according to image category. (b) The relative difference in contrast between bCFS and reCFS thresholds is the same across image categories.

Figure 4 with 1 supplement
Suppression depth is greater with less time for adaptation.

(a) bCFS (blue) and reCFS (red) thresholds collected during tCFS with three rates of contrast change (RCC) for the sample of 17 participants. Figure elements are the same as in Figure 3a. (b) Suppression depth increases when the rate of contrast change increases during tCFS. All error bars correspond to ± 1 SEM corrected for within participant comparisons (Cousineau, 2005). (c) Example trials at each rate of contrast change from a single participant. Red markers indicate reCFS responses, blue markers show bCFS.

Figure 4—figure supplement 1
Perceptual durations in Experiments 1–3.

(a) Average histogram across participants for all percept durations by experimental condition. Solid lines show tCFS conditions, broken lines show discrete (unidirectional) conditions. Shading corresponds to ± 1 SEM corrected for within participant comparisons (Cousineau, 2005). (b–c) Average histograms for tCFS in Experiment 2, and Experiment 3, respectively. Figure conventions as in (a). (d) tCFS durations of Experiment 3 when split by rate of contrast change. Vertical lines and arrows indicate the mean of median percept durations across participants.

Suppression depth over time is well described by a damped harmonic oscillator.

(a) Top row. Mean bCFS (blue) and reCFS (red) thresholds averaged by threshold order over the trial. Each column displays the average for slow, medium and fast rate of contrast change (RCC), respectively. (b) Middle row. Absolute change in target contrast between successive thresholds (i.e. suppression depth). Red bars indicate the decrease in target contrast needed for a visible target to reenter CFS. Blue bars indicate the increase in contrast for an invisible target to break CFS. (c) Bottom row. Yellow lines and shading plot the average change in suppression depth between successive thresholds. The data in each RCC condition is best fit by a damped harmonic oscillator (DHO) model shown in black. The change in Bayesian Information Criterion relative to the next best (cubic polynomial) model is displayed in each panel. All error bars and shading denote ± 1 SEM corrected for within participant comparisons (Cousineau, 2005).

Appendix 1—figure 1
Contrast response functions and CFS thresholds.

Two Naka-Rushton functions modelling contrast response functions (CRFs) are shown, one has parameters typical of primary visual cortical neurons (blue) and the other is typical of CRFs found in area V4 (red). The C50 and exponent values are taken from single-unit neurophysiological studies of primate areas V1 and V4. We make the simple assumption that when contrast is ramping up, a target will break through from suppression to visibility when response level reaches the ‘bCFS threshold’. Conversely, a visible target ramping down in contrast will become suppressed at the ‘reCFS threshold’. Note that the V4 CRF has bCFS/reCFS thresholds that are much lower than those from the V1 CRF, and that the bCFS/reCFS range is greater for the V4 CRF, implying greater suppression depth (see appendix text). Our finding of constant suppression depth and very similar bCFS/reCFS thresholds across diverse image types suggests that CFS suppression for all our tested target images occurs in a common mechanism likely to be early in binocular processing.

Tables

Table 1
Summary of results from model fits and comparisons.
Slow RCCMedium RCCFast RCC
ModelMeasure
(cubic) 1R2.07.19.29
BIC–774.311343.79–944.23
(harmonic) 2R2.005.0060
BIC–706.851138.595–600.34
(dHO) 3R2.684.632.361
BIC–1847.19–2125.29–1041.29
Comparison
2–1ΔBIC67.46205.195343.89
3–1ΔBIC–1072.88–781.5–97.06
3–2ΔBIC–1140.34–986.695–440.95

Additional files

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. David Alais
  2. Jacob Coorey
  3. Randolph Blake
  4. Matthew J Davidson
(2024)
A new ‘CFS tracking’ paradigm reveals uniform suppression depth regardless of target complexity or salience
eLife 12:RP91019.
https://doi.org/10.7554/eLife.91019.4