Attractor Ruins in Neural & Coupled Maps

• Attractor ruins are dynamical constructs that persist as residual organizing structures after traditional fixed-point stability is lost.
• In neural systems, slow adaptation of intrinsic parameters converts Hopfield attractors into transient ruins, enabling autonomous latching dynamics.
• In coupled map models, Milnor attractors lose strict asymptotic stability while retaining positive basins, leading to chaotic itinerancy among structured states.

Searching arXiv for papers on attractor ruins and related chaotic itinerancy. Attractor ruins, also termed attractor relicts, are dynamical structures obtained when attractors lose strict stability but continue to organize trajectories transiently or over a basin of positive measure. In a continuous neural-network setting, coupling a Hopfield-type attractor network to slow adaptation of intrinsic parameters destabilizes all fixed-point attractors and converts them into transiently visited “ruins,” yielding autonomous latching dynamics rather than convergence to a stationary state (Linkerhand et al., 2012). In a high-dimensional discrete setting, notably globally coupled logistic maps, attractor-ruins are described as Milnor attractors that have been “ruined,” i.e. they have lost strict asymptotic stability while retaining a nonzero measure basin; chaotic itinerancy then appears as motion among such quasi-stable cluster manifolds (Wada et al., 1 Oct 2025).

1. Definition and conceptual scope

In the standard attractor-network formulation, the dynamics descends in an energy or Lyapunov functional, and stable fixed points are minima of that functional. Once the state enters such an attractor, it remains there. A canonical example is the Hopfield energy

$$E(s)= -\frac12\sum_{i,j}w_{ij}s_i s_j + \sum_i \theta_i s_i,$$

or, for graded units and leaky integrators,

$$E[x] = -\tfrac12\sum_{i,j}(y_i w_{ij}y_j)^2 + \tfrac{\Gamma}{2}\sum_i x_i^2.$$

The attractor-ruin construction begins from this attractor-based organization but augments each neuron with slow adaptive intrinsic parameters such as thresholds or gains. Because the slow adaptation is driven by an information-theoretic objective that is incompatible with stationary firing-rate distributions, the original fixed points lose strict stability and survive only as transient slow regions in phase space (Linkerhand et al., 2012).

In the globally coupled map formulation, attractor-ruins are introduced through invariant cluster manifolds. For a permutation $\sigma \in S_N$,

$$H_\sigma := \{ x = (x(1),\dots,x(N)) \in [0,1]^N \mid x(i)=x(\sigma(i)) \text{ for all } i=1,\dots,N \}.$$

If the Lebesgue-measure basin

$$B(H_\sigma) := \{ x \in [0,1]^N \mid \mathrm{dist}(F^n(x),H_\sigma)\to 0 \text{ as } n\to\infty \}$$

has positive volume, then $H_\sigma$ is a Milnor attractor. If, under a small change of parameters, $H_\sigma$ ceases to be asymptotically stable but still attracts a positive-measure set of orbits, it is called an attractor-ruin (Wada et al., 1 Oct 2025).

A common misconception is to treat an attractor ruin as the disappearance of dynamical organization. In both formulations, the opposite is true: the system remains strongly structured, but the structure no longer supports permanent resting states. This suggests a unifying interpretation in which “ruin” denotes persistence without strict asymptotic trapping.

2. Generating-function construction in neural systems

The neural attractor-relict framework is built from two generating functionals. The first is the Hopfield energy, which generates a neural attractor network. The second is polyhomeostatic optimization, an information-theoretic generating functional encoding the information content of neural firing statistics (Linkerhand et al., 2012).

A convenient continuous-time formulation uses membrane potentials $x_i(t)$ and a sigmoidal transfer function

$y_i = g_i(x_i) = \frac{1}{1+\exp(a_i(b_i-x_i))},$

where $a_i$ is the gain and $$E[x] = -\tfrac12\sum_{i,j}(y_i w_{ij}y_j)^2 + \tfrac{\Gamma}{2}\sum_i x_i^2.$$0 the threshold. Minimizing the attractor-generating functional yields the fast dynamics

$$E[x] = -\tfrac12\sum_{i,j}(y_i w_{ij}y_j)^2 + \tfrac{\Gamma}{2}\sum_i x_i^2.$$1

With fixed $$E[x] = -\tfrac12\sum_{i,j}(y_i w_{ij}y_j)^2 + \tfrac{\Gamma}{2}\sum_i x_i^2.$$2, $$E[x] = -\tfrac12\sum_{i,j}(y_i w_{ij}y_j)^2 + \tfrac{\Gamma}{2}\sum_i x_i^2.$$3, and symmetric weights $$E[x] = -\tfrac12\sum_{i,j}(y_i w_{ij}y_j)^2 + \tfrac{\Gamma}{2}\sum_i x_i^2.$$4, the fixed points of these equations correspond to minima of the Hopfield energy.

The second functional specifies a target long-term firing-rate distribution $$E[x] = -\tfrac12\sum_{i,j}(y_i w_{ij}y_j)^2 + \tfrac{\Gamma}{2}\sum_i x_i^2.$$5 and measures mismatch via the Kullback–Leibler divergence

$$E[x] = -\tfrac12\sum_{i,j}(y_i w_{ij}y_j)^2 + \tfrac{\Gamma}{2}\sum_i x_i^2.$$6

where

$$E[x] = -\tfrac12\sum_{i,j}(y_i w_{ij}y_j)^2 + \tfrac{\Gamma}{2}\sum_i x_i^2.$$7

is the empirical firing-rate distribution of neuron $$E[x] = -\tfrac12\sum_{i,j}(y_i w_{ij}y_j)^2 + \tfrac{\Gamma}{2}\sum_i x_i^2.$$8. The target distribution is chosen as

$$E[x] = -\tfrac12\sum_{i,j}(y_i w_{ij}y_j)^2 + \tfrac{\Gamma}{2}\sum_i x_i^2.$$9

which fixes mean and variance. Stochastic gradient descent on $\sigma \in S_N$0 with respect to the intrinsic parameters produces the polyhomeostatic adaptation rules

$\sigma \in S_N$1

$\sigma \in S_N$2

with slow rates $\sigma \in S_N$3.

The technical significance of this construction is that the two functionals do not simply coexist; they compete. The Hopfield functional favors stationary pattern retrieval, whereas the polyhomeostatic functional favors firing statistics close to a broad target distribution. The resulting fast–slow system is therefore organized by incompatible optimization pressures.

3. Fast–slow destabilization and latching dynamics

Collecting the two generating functionals yields the coupled system

$\sigma \in S_N$4

together with the slow adaptation equations for $\sigma \in S_N$5 and $\sigma \in S_N$6. Because stationary firing patterns $\sigma \in S_N$7 are highly non-entropic, the Kullback–Leibler objective remains large and the adaptation gradients do not vanish. Consequently, no fixed point of the fast Hopfield subsystem survives indefinitely; every original attractor is continuously “chased away” by the slow evolution of intrinsic parameters (Linkerhand et al., 2012).

The resulting dynamics is latching. Former attractors persist only as transient slowdowns, and the trajectory moves from one such ruin to another. Autonomous dynamics involving sequences of transiently stable states have been termed associative latching in the context of grammar generation, and the same mechanism is proposed as relevant for motor control and associative neural computation in the brain. In larger Hopfield-encoded networks with $\sigma \in S_N$8 and stored patterns $\sigma \in S_N$9, each encoded pattern becomes an unstable slow region in phase space, and the trajectory visits these regions sequentially.

A minimal three-neuron example makes the mechanism explicit. For a symmetric network with one central neuron excitatory to two side neurons and mutual inhibition between the side neurons, the fixed-parameter system has a two-attractor region in the $$H_\sigma := \{ x = (x(1),\dots,x(N)) \in [0,1]^N \mid x(i)=x(\sigma(i)) \text{ for all } i=1,\dots,N \}.$$0 plane. Its lower boundary is a supercritical, second-order bifurcation,

$$H_\sigma := \{ x = (x(1),\dots,x(N)) \in [0,1]^N \mid x(i)=x(\sigma(i)) \text{ for all } i=1,\dots,N \}.$$1

where $$H_\sigma := \{ x = (x(1),\dots,x(N)) \in [0,1]^N \mid x(i)=x(\sigma(i)) \text{ for all } i=1,\dots,N \}.$$2 is the firing rate of the active side-sites. Since $$H_\sigma := \{ x = (x(1),\dots,x(N)) \in [0,1]^N \mid x(i)=x(\sigma(i)) \text{ for all } i=1,\dots,N \}.$$3, the critical gain is $$H_\sigma := \{ x = (x(1),\dots,x(N)) \in [0,1]^N \mid x(i)=x(\sigma(i)) \text{ for all } i=1,\dots,N \}.$$4. The upper boundary is a fold, or first-order, bifurcation. Under slow adaptation, the trajectories of $$H_\sigma := \{ x = (x(1),\dots,x(N)) \in [0,1]^N \mid x(i)=x(\sigma(i)) \text{ for all } i=1,\dots,N \}.$$5 numerically “sit” near the second-order boundary and repeatedly cross it, alternately destabilizing and re-stabilizing the former attractors.

This phase-boundary picture is important because it identifies attractor ruins not as an ad hoc perturbation effect but as a systematically maintained near-bifurcation regime.

4. Objective-function stress and temporal organization

The neural construction distinguishes two forms of mismatch between the generating functionals. The first is functional stress: Hopfield attractors imply sharply peaked firing-rate distributions, whereas polyhomeostatic adaptation demands a broad target distribution. This irreconcilable aim destroys the fixed points. The second is scalar, or mean-rate, stress: the Hopfield patterns have mean activity

$$H_\sigma := \{ x = (x(1),\dots,x(N)) \in [0,1]^N \mid x(i)=x(\sigma(i)) \text{ for all } i=1,\dots,N \}.$$6

while the target distribution has mean

$$H_\sigma := \{ x = (x(1),\dots,x(N)) \in [0,1]^N \mid x(i)=x(\sigma(i)) \text{ for all } i=1,\dots,N \}.$$7

When $$H_\sigma := \{ x = (x(1),\dots,x(N)) \in [0,1]^N \mid x(i)=x(\sigma(i)) \text{ for all } i=1,\dots,N \}.$$8, the network must either reduce or boost its firing rate relative to the stored-pattern level in order to optimize $$H_\sigma := \{ x = (x(1),\dots,x(N)) \in [0,1]^N \mid x(i)=x(\sigma(i)) \text{ for all } i=1,\dots,N \}.$$9 (Linkerhand et al., 2012).

If $$B(H_\sigma) := \{ x \in [0,1]^N \mid \mathrm{dist}(F^n(x),H_\sigma)\to 0 \text{ as } n\to\infty \}$$0, objective-function stress is absent at the level of mean activity. The slow adaptation then drives the system onto a near-periodic cycle that visits every attractor relict in turn, yielding regular latching with constant burst intervals. If $$B(H_\sigma) := \{ x \in [0,1]^N \mid \mathrm{dist}(F^n(x),H_\sigma)\to 0 \text{ as } n\to\infty \}$$1, the dynamics separates into long laminar phases far from any attractor relict and short bursting episodes of rapid latching among attractor relicts. For $$B(H_\sigma) := \{ x \in [0,1]^N \mid \mathrm{dist}(F^n(x),H_\sigma)\to 0 \text{ as } n\to\infty \}$$2, these bursts transiently raise the average activity; for $$B(H_\sigma) := \{ x \in [0,1]^N \mid \mathrm{dist}(F^n(x),H_\sigma)\to 0 \text{ as } n\to\infty \}$$3, they lower it. The observed regime is intermittent bursting latching with irregular intervals.

This distinction is not merely phenomenological. It ties a precise scalar mismatch between target statistics and encoded patterns to a qualitative change in temporal organization. A plausible implication is that attractor ruins can act as mediators between memory structure and homeostatic constraints, rather than as residual artifacts of destabilization.

5. Attractor-ruins in globally coupled maps

In the globally coupled map setting, the system is the $$B(H_\sigma) := \{ x \in [0,1]^N \mid \mathrm{dist}(F^n(x),H_\sigma)\to 0 \text{ as } n\to\infty \}$$4-dimensional map

$$B(H_\sigma) := \{ x \in [0,1]^N \mid \mathrm{dist}(F^n(x),H_\sigma)\to 0 \text{ as } n\to\infty \}$$5

with

$$B(H_\sigma) := \{ x \in [0,1]^N \mid \mathrm{dist}(F^n(x),H_\sigma)\to 0 \text{ as } n\to\infty \}$$6

For fixed $$B(H_\sigma) := \{ x \in [0,1]^N \mid \mathrm{dist}(F^n(x),H_\sigma)\to 0 \text{ as } n\to\infty \}$$7, orbits often collapse into clusters defined by some permutation $$B(H_\sigma) := \{ x \in [0,1]^N \mid \mathrm{dist}(F^n(x),H_\sigma)\to 0 \text{ as } n\to\infty \}$$8, meaning that they approach one of the subspaces $$B(H_\sigma) := \{ x \in [0,1]^N \mid \mathrm{dist}(F^n(x),H_\sigma)\to 0 \text{ as } n\to\infty \}$$9. These cluster manifolds act as attractors, or Milnor attractors, for many initial conditions (Wada et al., 1 Oct 2025).

The long-time behavior exhibits four characteristic phases in the $H_\sigma$0 plane: coherent, ordered, partially ordered, and turbulent. In the coherent phase, all components synchronize into a single cluster of dimension $H_\sigma$1. In the ordered phase, a small finite number of clusters forms and remains fixed. In the turbulent phase, the components remain desynchronized. The partially ordered phase is the regime in which the orbit does not settle on any single $H_\sigma$2 but wanders intermittently among a family of cluster manifolds.

It is precisely this partially ordered region that displays chaotic itinerancy. Orbits spend long epochs near one cluster manifold, undergo bursts of desynchronization, then settle near another cluster manifold. In this formulation, chaotic itinerancy is described directly as itineration among attractor-ruins. The characterization differs from the neural fast–slow construction, but the phenomenology is closely aligned: transient residence, loss of strict stability, and recurrent migration among structured states.

6. Quantification, simulation regimes, and interpretive issues

The globally coupled map study introduces two complementary quantitative measures. The first is the optimal-transport distance (OTD) between consecutive clustering patterns. At each time $H_\sigma$3, a clustering pattern $H_\sigma$4 with $H_\sigma$5 is mapped to a discrete distribution $H_\sigma$6 via

$H_\sigma$7

so that $H_\sigma$8 is the fraction of points that form clusters of size $H_\sigma$9. Given distributions $H_\sigma$0 and $H_\sigma$1 and a cost matrix $H_\sigma$2, the OTD is

$H_\sigma$3

subject to the transport constraints

$H_\sigma$4

When the clustering pattern has converged, OTD is identically zero after a transient; in the partially ordered phase it continues to fluctuate. The second measure is the Shannon-entropy strength of attractor-ruins,

$H_\sigma$5

where $H_\sigma$6 is the empirical distribution of effective dimensions $H_\sigma$7 at times satisfying $H_\sigma$8. Higher $H_\sigma$9 indicates that more types of cluster manifolds are visited, and more evenly visited, during itinerant motion. Both $x_i(t)$0 and $x_i(t)$1 are approximately zero in the coherent, ordered, and turbulent phases, and both rise in the partially ordered region, with $x_i(t)$2 peaking in “Partially Ordered II” (Wada et al., 1 Oct 2025).

The neural study reports representative simulation regimes with $x_i(t)$3 and $x_i(t)$4 up to $x_i(t)$5, leak rate $x_i(t)$6, Hopfield weights

$x_i(t)$7

pattern sparseness $x_i(t)$8, and $x_i(t)$9 to avoid severe interference. Typical adaptation rates are $y_i = g_i(x_i) = \frac{1}{1+\exp(a_i(b_i-x_i))},$0 and $y_i = g_i(x_i) = \frac{1}{1+\exp(a_i(b_i-x_i))},$1, and numerical integration uses 4th-order Runge–Kutta with $y_i = g_i(x_i) = \frac{1}{1+\exp(a_i(b_i-x_i))},$2. For $y_i = g_i(x_i) = \frac{1}{1+\exp(a_i(b_i-x_i))},$3, overlap traces $y_i = g_i(x_i) = \frac{1}{1+\exp(a_i(b_i-x_i))},$4 or $y_i = g_i(x_i) = \frac{1}{1+\exp(a_i(b_i-x_i))},$5 show a clean, regular latching cycle through all patterns; for $y_i = g_i(x_i) = \frac{1}{1+\exp(a_i(b_i-x_i))},$6, long low-activity laminar epochs are interrupted by fast bursts of pattern visits (Linkerhand et al., 2012).

Two interpretive cautions follow directly from these results. First, attractor ruins are not synonymous with noise-driven switching: in the neural construction they are generated by deterministic coupling of fast attractor dynamics to slow polyhomeostatic adaptation, and in the globally coupled map they are organized by cluster-manifold geometry and parameter-dependent loss of asymptotic stability. Second, attractor ruins are not equivalent to generic disorder. In the neural case, trajectories remain shaped by encoded Hopfield patterns; in the globally coupled map, coherent, ordered, and turbulent phases are all distinguished from the partially ordered itinerant regime by the behavior of OTD and $y_i = g_i(x_i) = \frac{1}{1+\exp(a_i(b_i-x_i))},$7. The available evidence therefore supports a narrow technical usage: an attractor ruin is a residual organizing structure that persists after strict attractor stability has been lost.