A Minimal Quantitative Model of Perceptual Suppression and Breakthrough in Visual Rivalry

arXiv:2510.17154 · q-bio.NC · Submitted 2025-10-20 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Genomics Radio. Generated commentary on the latest computational biology and genomics papers.

Ines: We're looking at the paper: "A Minimal Quantitative Model of Perceptual Suppression and Breakthrough in Visual Rivalry".

Marcus: Here's the gist of the paper. When conflicting images are presented to either eye binocular fusion is disrupted Rather than experiencing a blend of both percepts, often only one eye’s image is experienced, whilst the other is suppressed from awareness Suppression is transient – the two rival images compete for dominance, with stochastic switches between mutually exclusive percepts occurring every few seconds with law-like regularity From the perspective of dynamical systems theory, visual rivalry offers an experimentally tractable window into the dynamical mechanisms governing perceptual awareness In a recently developed visual rivalry paradigm – tracking continuous flash suppression (tCFS) – it was shown that the transition between awareness and suppression is hysteretic, with a higher contrast threshold required for a stimulus to breakthrough suppression into awareness than to be suppressed from awareness Here, we present an analytically-tractable model of visual rivalry that quantitatively explains the hysteretic transition between periods of awareness and suppression in tCFS Grounded in the theory of neural dynamics, we derive closed-form expressions for the duration of perceptual dominance and suppression, and for the degree of hysteresis (i.e., the depth of perceptual suppression), as a function of model parameters Finally, our model yields a series of novel behavioural predictions, the first of which – distributions of dominance and suppression durations during tCFS should be approximately equal - we empirically validate in human psychophysical data.

Ines: Model Development and Dynamics The dynamics of the model are described by a system of four nonlinear ordinary differential equations (ODEs) with reciprocal inhibition between competing monocular populations; selfadaptation; and percept-dependent stimulus contrast ramping These equations involve state variables representing the aggregate excitatory neural activity of orientation-selective monocular populations driven by the mask and stimulus (denoted by the subscripts M and S respectively), and the aggregate slow hyperpolarising adaptation current of each population.

Marcus: The model incorporates reciprocal inhibition between competing monocular populations; selfadaptation; and percept-dependent stimulus contrast ramping For analytic tractability, the transfer function is a simple threshold nonlinearity f(x) = max(x, 0).

Ines: The parameters ε, a, and g are dimensionless and represent the strength of recurrent excitation, competitive inhibition, and adaptation, respectively.

Title and authors: Marcus: The title itself, "A Minimal Quantitative Model of Perceptual Suppression and Breakthrough in Visual Rivalry," suggests they are aiming for something foundational rather than just adding another layer to existing theories.

Ines: They're trying to create the simplest possible mathematical structure that still captures the key features we see in these rivalry experiments, like when a stimulus finally breaks through suppression into awareness.

Yuki: I wonder what that "minimal" model actually means in practice, does it simplify the biology too much to be useful, or is it just focusing on the core competition mechanism?

Marcus: It seems they are using continuous flash suppression tracking, tCFS, as their setup because it captures those periods of suppression and awareness during a dynamic contrast situation.

Ines: In tCFS, one eye gets a high-contrast mask while the target stimulus in the other eye changes its contrast linearly based on what you report seeing. This setup allows them to measure thresholds for both breakthrough and suppression.

Yuki: So they are using this continuous dynamic stimulus rather than static images to probe these dynamical mechanisms, which is a big step because static rivalry is already well understood in some ways.

Marcus: The paper shows that the estimates they get at those points of suppression and breakthrough actually serve as threshold estimates for both awareness and suppression. This is useful for quantifying the dynamics.

Ines: They also mention that they make these estimates quickly so that the effect of adaptation is reduced, which helps keep the model tractable. This is a key engineering choice in their setup to isolate other effects.

Yuki: So they are trying to find a way to measure those thresholds without the system getting too bogged down by slow biological processes, like adaptation.

The paper's summary: Ines: What the paper really summarizes is that their model quantitatively explains the hysteretic transition between periods of awareness and suppression in tCFS. This means there's a difference between what it takes to break out of suppression versus what it takes to get suppressed in the first place.

Marcus: Hysteresis is a crucial finding here because it means the contrast threshold for breakthrough isn't the same as the threshold for being suppressed; you have different barriers depending on your perceptual history.

Yuki: That sounds like a real-world consequence, because it implies that how you've experienced these stimuli before shapes how you react to new ones.

Ines: Right, and they show that this degree of hysteresis depends on the rate of contrast change; faster changes in contrast lead to bigger differences between those two thresholds.

Marcus: They derive closed-form expressions for both the duration of dominance and suppression, and also for the depth of this hysteresis as a function of their model parameters like recurrent excitation strength, competitive inhibition, and adaptation.

Yuki: The idea that the depth of hysteresis is related to how long you've been in either state gives a nice link between dynamics and history.

Ines: They even find a symmetry in the results: the distributions for dominance and suppression durations should be approximately equal, which they have validated against human psychophysical data.

The paper's improvements: Marcus: One of the major improvements suggested by this work is the derivation of closed-form expressions for dominance and suppression duration that are written in terms of the Lambert W function, or product log function.

Ines: That means they can predict how long a state lasts just by plugging in the model parameters, instead of running a full simulation every single time. That’s really powerful for testing hypotheses computationally.

Yuki: It also suggests that these exact solutions provide an excellent fit to the durations derived from numerical simulations across the entire range of contrast rates, which validates their mathematical approach against other methods.

Marcus: They also use these expressions to show a trade-off between the opposing effects of contrast rate on those durations; increasing the rate shortens time but increases the gap between breakthrough and suppression thresholds.

Ines: So they're using this math to show that adaptation has an additive effect on breakthrough threshold but a subtractive effect on suppression threshold, and these two effects work against each other.

Yuki: That’s a neat way to quantify how different dynamical factors—the stationary competition versus the time-dependent adaptation—interact in shaping the perceptual experience.

Conclusion: Marcus: So, to wrap up, this paper provides a quantitative account of awareness and suppression in tCFS by deriving closed-form expressions for dominance and suppression periods and the depth of hysteresis between breakthrough and suppression thresholds.

Ines: The main contribution is explaining this effect quantitatively by framing it in terms of the trade-off between stationary effects from inhibition and external drive, versus time-dependent effects from adaptation.

Yuki: It confirms that these dynamical principles are likely going to be relevant for other rivalry paradigms beyond just tCFS, giving insight into a much wider range of how competing perceptions might arise.

Marcus: And they validated one of their novel predictions: the symmetry between periods of dominance and suppression durations is approximately equal, which they found in human psychophysical data.

Ines: So, this work gives us a quantitative tool to understand the mechanisms governing awareness and suppression in continuous flash suppression, moving beyond just observing the phenomena.

Centre for Complex Systems, University of Sydney, Australia · Neuroscience Research Theme, School of Medical Sciences, Faculty of Medicine and Health, University of Sydney · Centre for Vision Research, York University, Toronto · School of Mathematical and Physical Sciences, Macquarie University · Computational Neuroscience Department at Technische Universität Dresden · Department of Computational Neuroscience at Max Planck Institute for Biological Cybernetics · Discipline of Psychology Graduate School of Health, University Technology Sydney · School of Psychology, University of Sydney

q-bio.NC

Submitted: 2025-10-20

Updated: 2026-11-13

Code: https://github.com/cjwhyte/tCFS

License: http://creativecommons.org/licenses/by-nc-sa/4.0/

Importance score: 94/100

The gist: When conflicting images are presented to either eye binocular fusion is disrupted Rather than experiencing a blend of both percepts, often only one eye’s image is experienced…

Key concepts

Visual Rivalry
This occurs when two conflicting images are presented to both eyes simultaneously, disrupting binocular fusion. Instead of seeing a blend, one image dominates while the other is suppressed. This competition between rival images is modeled using dynamical systems theory.
Hysteresis
Hysteresis refers to the dependence of a system's state on its history. In this context, it means that the threshold required for a stimulus to break through suppression into awareness (breakthrough) is different from the threshold needed to be suppressed from awareness. This difference is quantified by the depth of suppression.
Continuous Flash Suppression (tCFS)
This paradigm involves tracking how a stimulus's dominance and suppression change continuously over time. The model analyzes how the transition between awareness and suppression behaves under these non-stationary, rapidly changing visual conditions, revealing its underlying dynamical mechanisms.

Terminology

Summary

When conflicting images are presented to either eye binocular fusion is disrupted Rather than experiencing a blend of both percepts, often only one eye’s image is experienced, whilst the other is suppressed from awareness Suppression is transient – the two rival images compete for dominance, with stochastic switches between mutually exclusive percepts occurring every few seconds with law-like regularity From the perspective of dynamical systems theory, visual rivalry offers an experimentally tractable window into the dynamical mechanisms governing perceptual awareness In a recently developed visual rivalry paradigm – tracking continuous flash suppression (tCFS) – it was shown that the transition between awareness and suppression is hysteretic, with a higher contrast threshold required for a stimulus to breakthrough suppression into awareness than to be suppressed from awareness Here, we present an analytically-tractable model of visual rivalry that quantitatively explains the hysteretic transition between periods of awareness and suppression in tCFS Grounded in the theory of neural dynamics, we derive closed-form expressions for the duration of perceptual dominance and suppression, and for the degree of hysteresis (i. p. 0

Where the paper says it

Importantly, suppression is transient – the two rival images compete for dominance, with stochastic switches between mutually exclusive percepts occurring every few seconds with law-like regularity. From the perspective of dynamical systems theory, visual rivalry offers an experimentally tractable window into the dynamical mechanisms governing perceptual awareness.

e., the depth of perceptual suppression), as a function of model parameters Finally, our model yields a series of novel behavioural predictions, the first of which – distributions of dominance and suppression durations during tCFS should be approximately equal - we empirically validate in human psychophysical data. p. 0
Where the paper says it

, the depth of perceptual suppression), as a function of model parameters. Finally, our model yields a series of novel behavioural predictions, the first of which – distributions of dominance and suppression durations during tCFS should be approximately equal - we empirically validate in human psychophysical data.

Model Development and Dynamics

The dynamics of the model are described by a system of four nonlinear ordinary differential equations (ODEs) with reciprocal inhibition between competing monocular populations; selfadaptation; and percept-dependent stimulus contrast ramping These equations involve state variables representing the aggregate excitatory neural activity of orientation-selective monocular populations driven by the mask and stimulus (denoted by the subscripts M and S respectively), and the aggregate slow hyperpolarising adaptation current of each population. The model incorporates reciprocal inhibition between competing monocular populations; selfadaptation; and percept-dependent stimulus contrast ramping For analytic tractability, the transfer function is a simple threshold nonlinearity f(x) = max(x, 0). p. 2

Where the paper says it

Finally, we derive two additional behavioural predictions that can be tested in future experiments. 2 Model The dynamics of the model are described by a system of four nonlinear ordinary differential equations (ODEs) with: reciprocal inhibition between competing monocular populations; selfadaptation; and percept-dependent stimulus contrast ramping (Figure 1A).

The parameters ε, a, and g are dimensionless and represent the strength of recurrent excitation, competitive inhibition, and adaptation, respectively. p. 3
Where the paper says it

simple threshold nonlinearity f(x) = max(x, 0). The parameters ε, a, and g, are dimensionless and represent the strength of recurrent excitation, competitive inhibition, and adaptation, respectively.

Hysteresis Quantification

Two key empirical findings emerge from this paradigm:

  1. The contrast threshold for breakthrough and suppression is hysteretic across three experiments with multiple stimulus categories This means the same stimulus, with the same contrast could dominant the content of awareness, or be suppressed from awareness, depending on the participant’s history of perceptual stimulation. p. 6

    Where the paper says it

    Across three experiments with multiple stimulus categories, the average difference between breakthrough and suppression thresholds was shown to be constant, with a reliably higher threshold for breakthrough than suppression. That is, the same stimulus, with the same contrast could dominant the content of awareness, or be suppressed from awareness, depending on the participant’s history of perceptual stimulation.

  2. The degree of hysteresis is dependent on the rate of contrast change, with faster rates of contrast change leading to larger differences in breakthrough and suppression thresholds. p. 6

    Where the paper says it

    That is, the same stimulus, with the same contrast could dominant the content of awareness, or be suppressed from awareness, depending on the participant’s history of perceptual stimulation. Second, the degree of hysteresis is dependent on the rate of contrast change, with faster rates of contrast change leading to larger differences in breakthrough and suppression thresholds.

The depth of hysteresis is given by contrast rate and the length of the dominance/suppression period The analytical approach explains this trend quantitatively by deriving expressions for the value of the target stimulus at breakthrough (SB) and suppression (SS) points. p. 6

Where the paper says it

In line with empirical data, in our simulations, the depth of hysteresis in contrast thresholds increased as a linear function of contrast rate. With our analytical approach, we can explain this trend quantitatively by deriving expressions for the value of the target stimulus at breakthrough (SB) and suppression (SS) points.

Duration Predictions

The closed-form expressions for dominance and suppression duration can be seen as a solution to the trade-off between the opposing effects of contrast rate on dominance and suppression duration As contrast rate increases, it takes less time for the stimulus to change from SB to SS but by the same token, because adaptation has an additive effect on breakthrough threshold and a subtractive effect on suppression threshold, as dominance durations decrease, the distance between breakthrough and suppression values increases.

The expressions for dominance and suppression duration can be shown to have exact solutions written in terms of the Lambert W (i. p. 8

Where the paper says it

In essence, then, we can think of each successive period of dominance and suppression as setting the initial conditions for the following period. Although we cannot solve for t in terms of elementary functions, both expressions can be shown to have exact solutions written in terms of the Lambert W (i.

e. product log) function These closed-form expressions provide an excellent fit to dominance and suppression durations derived from numerical simulations across the full range of contrast rates. p. 8
Where the paper says it

Both equations provide an excellent fit to dominance and suppression durations derived from numerical simulations across the full range of contrast rates (Figure 3A). Full derivation of equation 10 is supplied in Appendix 2 The first term in both expressions captures the stationary contribution of the difference between the excitatory drive entering the suppressed population and the level of asymptotic inhibition from the dominant population.

Empirical Validation and Future Directions

The model accurately reproduces existing data and tested a novel prediction in existing data The symmetry of dominance and suppression durations is a prediction that we can test in the empirical data of Alais and colleagues We took a two-pronged approach to test this prediction, including using a two-sample Kolmogorov-Smirnov test against the null hypothesis that the data come from the same underlying distribution.

Future experiments can test predictions such as pharmacological manipulation of competitive inhibition should increase suppression depth and equalising the strength of adaptation between competing neural populations should approximately halve the depth of hysteresis between breakthrough and suppression thresholds. p. 2

Where the paper says it

Finally, we derive two additional behavioural predictions that can be tested in future experiments. 2 Model The dynamics of the model are described by a system of four nonlinear ordinary differential equations (ODEs) with: reciprocal inhibition between competing monocular populations; selfadaptation; and percept-dependent stimulus contrast ramping (Figure 1A).

The study suggests that these dynamical principles will likely go beyond tCFS, giving insight into a wide range of rivalry-like paradigms. p. 1
Where the paper says it

Closed-form expressions for each law can be derived from models with physiologically meaningful parameters [24, 30, 38], offering a quantitative explanation of the relative dominance of each percept in terms of the underlying dynamical processes and their coarse-grained physiological implementation. The minimal conditions for awareness and suppression in continuous flash suppression (CFS;[45]), another canonical rivalry paradigm, can also be understood in terms of the same underlying principles.

Data and Code Availability

Code to reproduce all simulations can be downloaded from https://github.com/cjwhyte/tCFS The empirical data from [47] analysed in the paper can be downloaded from the open science framework: https://osf. p. 4

Where the paper says it

Figure 1: A) tCFS model architecture – monocular populations driven by a mask (M) with constant contrast, and target stimulus (S) with percept-dependent ramping contrast, compete through mutual inhibition with asymptotic dynamics governed by adaptation. B) Dominance duration loss (∆model-data) quantifying the fit between empirical and simulated dominance durations across the [gM × M] parameter space.

io/nzp9v.

Appendix Details

The appendix provides the derivation of dominance and suppression duration formula, showing that both equations can be shown to have the form y = xex which for real x and y, on the restricted domain x ≥ 0, have an inverse function W0 (i.e. x = W0(xex)), where W0 is the principle branch of the Lambert W function. p. 8

Where the paper says it

| {z } Stationary drive + τHW0 gMM(1 + gS − ε) aγτH(1 + gM − ε) e − 1 γτH SB− (1+gS−ε) a M ! | {z } Time-dependent drive , where W0 is the principle branch of the Lambert W function.

The derivation of dominance time from line two of equation 9 is slightly more complex, as changes in the strength of the stimulus have a delayed effect on the subthreshold dynamics of the mask-driven population. p. 3
Where the paper says it

Unlike binocular rivalry, where the external drive shown to both eyes is stationary, in tCFS, a dynamic (flashing) mask is presented to one eye and the contrast of the target stimulus presented to the other eye changes (log) linearly with time. To approximate the effect of the dynamic mask whilst keeping the model analytically tractable, we drove the mask population with a constant stimulus but reduced the strength of adaptation in line with previous hypotheses [47] and modelling [46].

The iterative update algorithm allows for convergence to equilibrium values shown in Figure 3 and Figure 4. p. 9

Where the paper says it

Crucially, once the model has converged to an equilibrium, the expressions become interchangeable (i. HS→B = HB→S) and provide an excellent fit to the hysteresis depth values derived from numerical simulation (Figure 3B).

The difference equations rapidly converged across all contrast rates to the equilibrium values shown in Figure 3 and Figure 4. p. 8
Where the paper says it

Both equations provide an excellent fit to dominance and suppression durations derived from numerical simulations across the full range of contrast rates (Figure 3A). Full derivation of equation 10 is supplied in Appendix 2 The first term in both expressions captures the stationary contribution of the difference between the excitatory drive entering the suppressed population and the level of asymptotic inhibition from the dominant population.

The symmetry between periods of dominance and suppression is understood geometrically by treating each perceptual switch from dominance to suppression (and vice versa) as a bifurcation. p. 9

Where the paper says it

One can understand the symmetry between periods of dominance and suppression by noting that: SSn+1 = SBn − γTDominantn , (12) SBn+1 = SSn+1 + γTSuppressedn+1 , which, because of the successive dependence of breakthrough and suppression thresholds, leads to, SSn+1 = SSn + γTSuppressedn − γTDominantn , SBn+1 = SBn − γTDominantn + γTSuppressedn+1 , implying that once the model has converged to equilibrium, where SBn+1 = SBn and SSn+1 = SSn , it must be the case that TDominantn = TSuppressedn (Figure 3A-B). We can understand this symmetry geometrically by treating each perceptual switch from dominance to suppression (and vice versa) as a bifurcation and constructing bifurcation diagrams for each population with bifurcation points given by SB and SS (Figure 3C).

The distance between SB and SS is constant across periods of dominance and suppression (once the model has reached equilibrium), implying that TDominantn = TSuppressedn. p. 9
Where the paper says it

HS→B = HB→S) and provide an excellent fit to the hysteresis depth values derived from numerical simulation (Figure 3B). One can understand the symmetry between periods of dominance and suppression by noting that: SSn+1 = SBn − γTDominantn , (12) SBn+1 = SSn+1 + γTSuppressedn+1 , which, because of the successive dependence of breakthrough and suppression thresholds, leads to, SSn+1 = SSn + γTSuppressedn − γTDominantn , SBn+1 = SBn − γTDominantn + γTSuppressedn+1 , implying that once the model has converged to equilibrium, where SBn+1 = SBn and SSn+1 = SSn , it must be the case that TDominantn = TSuppressedn (Figure 3A-B).

The paper concludes by showing that the model extends insights from quantitative models of binocular rivalry [30], and CFS [46], to tCFS. p. 2

Where the paper says it

In a companion paper [48], we generalised a minimal model of binocular rivalry [30] to tCFS, and used numerical simulations to show that a singular mechanism, competitive inhibition between slowly adapting monocular populations, can account for perceptual awareness and suppression in both paradigms. Here, we use the same model to develop a quantitative account of the mechanisms governing awareness and suppression.

The key contribution is the derivation of closed-form expressions for the duration of dominance and suppression periods for the type of non-stationary stimuli used in tCFS, and an accompanying expression for the depth of hysteresis (i. p. 9
Where the paper says it

As contrast rate increases, it takes less time for the stimulus to change from the breakthrough value to the suppression value (and vice versa), but by the same token, because adaptation has an additive effect on the breakthrough threshold, and a subtractive effect on the suppression threshold (see equations 7-8), as dominance durations decrease, the distance between breakthrough and suppression values increases. Closed-form expressions for dominance and suppression duration in hand, we can now derive an expression for hysteresis depth as a function of contrast rate.

p. 9
Where the paper says it

As contrast rate increases, it takes less time for the stimulus to change from the breakthrough value to the suppression value (and vice versa), but by the same token, because adaptation has an additive effect on the breakthrough threshold, and a subtractive effect on the suppression threshold (see equations 7-8), as dominance durations decrease, the distance between breakthrough and suppression values increases. Closed-form expressions for dominance and suppression duration in hand, we can now derive an expression for hysteresis depth as a function of contrast rate.

e., suppression depth) between breakthrough and suppression thresholds. p. 2
Where the paper says it

The threshold reached in this way will be less affected by adaptation. Intriguingly, stimulus contrast thresholds measured at points of breakthrough and suppression show a robust pattern of hysteresis (known as suppression depth in the visual rivalry literature), with a higher contrast threshold required for a stimulus to breakthrough suppression into awareness, than to be suppressed from awareness.

p. 2
Where the paper says it

The threshold reached in this way will be less affected by adaptation. Intriguingly, stimulus contrast thresholds measured at points of breakthrough and suppression show a robust pattern of hysteresis (known as suppression depth in the visual rivalry literature), with a higher contrast threshold required for a stimulus to breakthrough suppression into awareness, than to be suppressed from awareness.

p. 2
Where the paper says it

The threshold reached in this way will be less affected by adaptation. Intriguingly, stimulus contrast thresholds measured at points of breakthrough and suppression show a robust pattern of hysteresis (known as suppression depth in the visual rivalry literature), with a higher contrast threshold required for a stimulus to breakthrough suppression into awareness, than to be suppressed from awareness.

The model explains this effect quantitatively in terms of the trade-off between the (relatively) stationary effects of inhibition and external drive, and the time-dependent effects of adaptation. p. 8
Where the paper says it

Full derivation of equation 10 is supplied in Appendix 2 The first term in both expressions captures the stationary contribution of the difference between the excitatory drive entering the suppressed population and the level of asymptotic inhibition from the dominant population. In fact, if we neglect the time-dependent effects of adaptation, we can derive the stationary component of equation 10 by substituting the exact solution for the stimulus dynamics into equations 7-8 and solving for t.

p. 8
Where the paper says it

Full derivation of equation 10 is supplied in Appendix 2 The first term in both expressions captures the stationary contribution of the difference between the excitatory drive entering the suppressed population and the level of asymptotic inhibition from the dominant population. In fact, if we neglect the time-dependent effects of adaptation, we can derive the stationary component of equation 10 by substituting the exact solution for the stimulus dynamics into equations 7-8 and solving for t.

p. 8
Where the paper says it

Full derivation of equation 10 is supplied in Appendix 2 The first term in both expressions captures the stationary contribution of the difference between the excitatory drive entering the suppressed population and the level of asymptotic inhibition from the dominant population. In fact, if we neglect the time-dependent effects of adaptation, we can derive the stationary component of equation 10 by substituting the exact solution for the stimulus dynamics into equations 7-8 and solving for t.

The study confirms that parameter estimates for dominance and suppression distributions were near identical for both gamma (αDom = 4. p. 0

Where the paper says it

, the depth of perceptual suppression), as a function of model parameters. Finally, our model yields a series of novel behavioural predictions, the first of which – distributions of dominance and suppression durations during tCFS should be approximately equal - we empirically validate in human psychophysical data.

64, θDom = 0.784; αSup = 4.63, θSup = 0.78) and lognormal (µDom = 1.18, σDom = 0.49; µSup = 1.17, σSup = 0.48) distributions with almost entirely overlapping 95% confidence intervals. p. 0
Where the paper says it

, the depth of perceptual suppression), as a function of model parameters. Finally, our model yields a series of novel behavioural predictions, the first of which – distributions of dominance and suppression durations during tCFS should be approximately equal - we empirically validate in human psychophysical data.

p. 0
Where the paper says it

, the depth of perceptual suppression), as a function of model parameters. Finally, our model yields a series of novel behavioural predictions, the first of which – distributions of dominance and suppression durations during tCFS should be approximately equal - we empirically validate in human psychophysical data.

The model provides a quantitative account of the mechanisms governing awareness and suppression in tCFS. p. 2

Where the paper says it

In a companion paper [48], we generalised a minimal model of binocular rivalry [30] to tCFS, and used numerical simulations to show that a singular mechanism, competitive inhibition between slowly adapting monocular populations, can account for perceptual awareness and suppression in both paradigms. Here, we use the same model to develop a quantitative account of the mechanisms governing awareness and suppression.

The key contribution is the derivation of closed-form expressions for the duration of dominance and suppression periods for the type of non-stationary stimuli used in tCFS, and an accompanying expression for the depth of hysteresis (i.e., suppression depth) between breakthrough and suppression thresholds. The model explains this effect quantitatively in terms of the trade-off between the (relatively) stationary effects of inhibition and external drive, and the time-dependent effects of adaptation.

The paper confirms that parameter estimates for dominance and suppression distributions were near identical for both gamma (αDom = 4. p. 0

Where the paper says it

, the depth of perceptual suppression), as a function of model parameters. Finally, our model yields a series of novel behavioural predictions, the first of which – distributions of dominance and suppression durations during tCFS should be approximately equal - we empirically validate in human psychophysical data.

64, θDom = 0.784; αSup = 4.63, θSup = 0.78) and lognormal (µDom = 1.18, σDom = 0.49; µSup = 1.17, σSup = 0.48) distributions with almost entirely overlapping 95% confidence intervals.

The model provides a quantitative account of the mechanisms governing awareness and suppression in tCFS The key contribution is the derivation of closed-form expressions for the duration of dominance and suppression periods for the type of non-stationary stimuli used in tCFS, and an accompanying expression for the depth of hysteresis (i. p. 2

Where the paper says it

Here, we use the same model to develop a quantitative account of the mechanisms governing awareness and suppression. Specifically, we extend the quantitative study of visual rivalry to the threshold for awareness and suppression by deriving closed-form expressions for dominance and suppression durations, and for the degree of hysteresis separating breakthrough and suppression thresholds, as a function of model parameters.

p. 2
Where the paper says it

Here, we use the same model to develop a quantitative account of the mechanisms governing awareness and suppression. Specifically, we extend the quantitative study of visual rivalry to the threshold for awareness and suppression by deriving closed-form expressions for dominance and suppression durations, and for the degree of hysteresis separating breakthrough and suppression thresholds, as a function of model parameters.

e., suppression depth) between breakthrough and suppression thresholds The model explains this effect quantitatively in terms of the trade-off between the (relatively) stationary effects of inhibition and external drive, and the time-dependent effects of adaptation. p. 9
Where the paper says it

The closed-form expressions for dominance and suppression duration can, therefore, be seen as a solution to the trade-off between the opposing effects of contrast rate on dominance and suppression duration. As contrast rate increases, it takes less time for the stimulus to change from the breakthrough value to the suppression value (and vice versa), but by the same token, because adaptation has an additive effect on the breakthrough threshold, and a subtractive effect on the suppression threshold (see equations 7-8), as dominance durations decrease, the distance between breakthrough and suppression values increases.

The model provides a quantitative account of the mechanisms governing awareness and suppression in tCFS The key contribution is the derivation of closed-form expressions for the duration of dominance and suppression periods for the type of non-stationary stimuli used in tCFS, and an accompanying expression for the depth of hysteresis (i.e., suppression depth) between breakthrough and suppression thresholds. The model explains this effect quantitatively in terms of the trade-off between the (relatively) stationary effects of inhibition and external drive, and the time-dependent effects of adaptation.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improved AI systems can quantitatively explain visual rivalry dynamics by deriving closed-form expressions for dominance and suppression durations in tracking continuous flash suppression (tCFS). This allows AI to predict empirically validated results, such as distributions of dominance and suppression durations during tCFS should be approximately equal.

  2. Improved AI systems can calculate the degree of hysteresis between perceptual thresholds as a function of contrast rate. This provides a quantitative account for how faster rates of contrast change leading to larger differences in breakthrough and suppression thresholds by deriving expressions for HS over time-dependent drive.

  3. Improved AI systems can generate novel behavioral predictions based on model parameters, such as the prediction that the distributions of dominance and suppression durations should be approximately equal, which has been empirically validated in human psychophysical data.

  4. Improved AI systems can test hypotheses regarding pharmacological manipulation by simulating how increasing competitive inhibition affects the depth of hysteresis, showing that increasing competitive inhibition increased the depth of hysteresis across the full range of contrast rates.

  5. Improved AI systems can assess the effect of manipulating adaptation strength by simulating tracking binocular rivalry (tBR), which shows that equalizing adaptation strength almost halves the depth of hysteresis seen in standard tCFS.

Abstract

When conflicting images are presented to either eye, binocular fusion is disrupted. Rather than experiencing a blend of both percepts, often only one eye's image is experienced, whilst the other is suppressed from awareness. Importantly, suppression is transient - the two rival images compete for dominance, with stochastic switches between mutually exclusive percepts occurring every few seconds with law-like regularity. From the perspective of dynamical systems theory, visual rivalry offers an experimentally tractable window into the dynamical mechanisms governing perceptual awareness. In a recently developed visual rivalry paradigm - tracking continuous flash suppression (tCFS) - it was shown that the transition between awareness and suppression is hysteretic, with a higher contrast threshold required for a stimulus to breakthrough suppression into awareness than to be suppressed from awareness. Here, we present an analytically-tractable model of visual rivalry that quantitatively explains the hysteretic transition between periods of awareness and suppression in tCFS. Grounded in the theory of neural dynamics, we derive closed-form expressions for the duration of perceptual dominance and suppression, and for the degree of hysteresis (i.e. the depth of perceptual suppression), as a function of model parameters. Finally, our model yields a series of novel behavioural predictions, the first of which - distributions of dominance and suppression durations during tCFS should be approximately equal - we empirically validate in human psychophysical data.

Related papers