Selective attention can momentarily alter visual appearance, but can such effects be learned? We tested whether training attention under sensory competition produces lasting changes in perceived contrast. Across seven days, participants trained on an orientation task with a fixed target location, with or without a salient distractor. Before and after training, we measured the point of subjective equality (PSE). Training under competition produced a reliable push-pull shift. Stimuli at the trained location appeared higher in contrast, whereas stimuli at the untrained location appeared lower. Conversely, training without distractors improved performance but did not alter appearance. Crucially, these opponent shifts were robust to task variations, persisting even in equality judgments designed to minimize response bias. Furthermore, the effect generalized to stimuli with novel orientation and contrast levels. These findings demonstrate that resolving sensory competition does not merely improve discrimination, but durably recalibrates the subjective appearance of the visual world.
Continual reinforcement learning aims to produce agents that learn not only to improve at their current tasks but also to adapt as task distributions change. Training an agent on many diverse tasks can induce zero-shot generalization, but previous work generally evaluates this generalization after training -- with frozen weights. Whether task diversity also improves an agent's ability to continue learning across distribution shifts remains unclear. We introduce Banyan, a GPU-accelerated continual RL domain in which task diversity factors into three independently controllable axes: the map layouts an agent must navigate, the objects it must interact with, and the hierarchical structures of sub-goal dependencies. Across individual distribution shifts, increasing diversity along each axis causes agents to begin training on the new tasks near the performance attained on the previous one, even when the shift changes the structure of the optimal policy. However, as the number of shifts increases, this local transfer does not by itself yield sustained continual learning: longer-horizon tasks plateau, and earlier task distributions are forgotten after later training. Banyan is a benchmark
Large language models (LLMs) are transforming web search by shifting from document ranking to synthesizing answers, and are increasingly deployed as autonomous agentic search systems that iteratively interact with external knowledge sources. Despite this progress, building effective search agents remains challenging because high-quality intermediate search steps are difficult to generate. Previous approaches have primarily relied on outcome supervision, rewarding agents only for producing correct final answers. This often leads to reward hacking and excessive dependence on parametric memory, limiting generalization to out-of-domain tasks. To address these limitations, we introduce RAG-Gym, a framework that shifts supervision from final answers to the search process itself. With RAG-Gym, we systematically investigate architecture design, parameter optimization, and action evaluation, identifying reasoning reflection as a critical capability for search agents. Building on this insight, we propose Re$^2$Search++, a process-supervised agent that achieves substantial improvements on multi-hop information-seeking benchmarks, especially in out-of-domain settings. Performance gains are dri
The on-off phenomena of opponent colors in center-surround may be the best-known facts of retinal processing of information. Apparently, however, no explicit model has been proposed that shows how neurons can be connected to produce the center-surround phenomena. Here it is shown that a previous simple model of color vision can produce these phenomena, including the detection of edge orientation and motion famously discovered by Hubel and Wiesel. The model was previously shown to produce major phenomena central to color vision, including mutually exclusive opponent colors. Although the opponencies of mutually exclusive colors and center-surround involve the same color pairs, red-green and blue-yellow, the model produces them by two different mechanisms. Perceptions of two colors are mutually exclusive because only one cone type can have the most, or least, absorption of photons. Two colors have the on-off opponency of center-surround because they have the same network designs up to the ganglion cells, with the inputs reversed there. On-off opponencies with different colors are possible, but natural selection evidently did not choose them.
The colourful wings of butterflies result from the interaction between light and the intricate chitinous nanostructures on butterflies' scales. This study demonstrates that just by reproducing the chitinous ridges present in butterfly scales (i.e., without any other secondary structure), the entire colour palette is achieved. This result was achieved using a new methodology based on the controlled reproduction of parts of the biological structure of complex chitinous systems using their native chemistry, enabling the isolation of different features' contributions. Here we isolate the contribution of the ridges and their variations as producing and modulating colour hue. The results suggest that complicated butterfly scales may be non-ideal solutions for producing colour when multifunctionality is not considered.
It is shown that for a generic electrovacuum spacetime, electromagnetic radiation produces vorticity of worldlines of observers in a Bondi--Sachs frame. Such an effect (and the ensuing gyroscope precession with respect to the lattice) which is a reminiscence of generation of vorticity by gravitational radiation, may be linked to the nonvanishing of components of the Poynting and the super--Poynting vectors on the planes othogonal to the vorticity vector. The possible observational relevance of such an effect is commented.
We discuss an algorithm which produces the meaning of a sentence given meanings of its words, and its resemblance to quantum teleportation. In fact, this protocol was the main source of inspiration for this algorithm which has many applications in the area of Natural Language Processing.
In this article we report that the entanglement produced by single photon subtraction is maximum, by studying the entanglement of multi photon subtraction from two mode squeezed states. We argue that the single photon subtraction produces maximum entanglement as it is more non-classical in nature than many photon subtractions
We study, for the first time, the spatial extension of the "source" that produces quark gluon plasma (QGP) in ultra relativistic heavy ion collisions. The longitudinal dimension is studied as a function of time as the system evolves. The source size is found to exhibit a novel non-classical feature.
Fracton systems exhibit restricted mobility of their excitations due to the presence of higher-order conservation laws. Here we study the time evolution of a one-dimensional fracton system with charge and dipole moment conservation using a random unitary circuit description. Previous work has shown that when the random unitary operators act on four or more sites, an arbitrary initial state eventually thermalizes via a universal subdiffusive dynamics. In contrast, a system evolving under three-site gates fails to thermalize due to strong "fragmentation" of the Hilbert space. Here we show that three-site gate dynamics causes a given initial state to evolve toward a highly nonthermal state on a time scale consistent with Brownian diffusion. Strikingly, the dynamics produces an effective attraction between isolated fractons or between a single fracton and the boundaries of the system, in analogy with the Casimir effect in quantum electrodynamics. We show how this attraction can be understood by exact mapping to a simple classical statistical mechanics problem, which we solve exactly for the case of an initial state with either one or two fractons.
We show that the ribbon zesting construction can produce modular isotopes -- different modular fusion categories with the same modular data. The result relies on the observation that the Reshetikhin-Turaev invariants of framed links associated to a ribbon fusion category satisfy a factorization property under zesting. This gives a new perspective on using topological invariants to classify topological order in light of modular data not being a complete invariant.
Data augmentation can mitigate limited training data in machine-learning automated scoring engines for constructed response items. This study seeks to determine how well three approaches to large language model prompting produce essays that preserve the writing quality of the original essays and produce realistic text for augmenting ASE training datasets. We created simulated versions of student essays, and human raters assigned scores to them and rated the realism of the generated text. The results of the study indicate that the predict next prompting strategy produces the highest level of agreement between human raters regarding simulated essay scores, predict next and sentence strategies best preserve the rated quality of the original essay in the simulated essays, and predict next and 25 examples strategies produce the most realistic text as judged by human raters.
This paper proposes a Situation Calculus solution to the frame problem for obligation-producing actions, which are actions that create obligations on the part of the agent that performs them. As an example of such actions, we have an opening door action performed by an agent, which has the subsequent obligation of getting the door closed. Demolombe and others extend Raymond Reiter's solution to the frame problem for ordinary actions to accommodate obligation-producing actions. Obligation-producing actions do affect the truth value of a newly introduced fluent that captures the accessibility relation used in semantics of obligation modalities in the Situation Calculus. Our work simplifies Demolombe's characterization of the accessibility relation by eliminating the notion of ideality of situations, thereby remaining close to Kripke-style possible-world semantics for deontic logic, in the spirit of Governatori's approach. Furthermore, we spell out details of a complete solution by extending basic action theories of Reiter to the new setting. Finally, we extend Reiter's regression operator for reasoning about actions back to the initial situation to this new setting. Our solution yiel
The article focuses on word (or string) attractors, which are sets of positions related to the text compression efficiency of the underlying word. The article presents two combinatorial algorithms based on Suffix automata or Directed Acyclic Word Graphs. The first algorithm decides in linear time whether a set of positions on the word is an attractor of the word. The second algorithm generates an attractor for a given word in a greedy manner. Although this problem is NP-hard, the algorithm is efficient and produces very small attractors for several well-known families of words.
Sufficiently strong electric fields can produce charged-particle pairs via the Schwinger effect. We argue that steep matter-density gradients, as can arise in neutron star interiors, would analogously produce neutrino-antineutrino pairs. We then discuss observational signatures of these gradient-produced (anti)neutrinos and how they could provide new probes of neutron-star structure and baryon-dense QCD.
We survey the known group properties that a sequence of finite groups or group actions needs to satisfy to admit subsets of bounded cardinality producing expander Cayley or Schreier graphs. We prove that an infinite amenable group and solvable groups of bounded derived length do not produce expander Schreier graphs, generalizing with easier proofs results of Lubotzky and Weiss for Cayley graphs. In particular, the poor expansion properties of a group action cannot in general be detected by looking at the abelian sections or at the representations above the stabilizer of a point.
Neural networks are typically black-boxes that remain opaque with regards to their decision mechanisms. Several works in the literature have proposed post-hoc explanation methods to alleviate this issue. This paper proposes LMAC-TD, a post-hoc explanation method that trains a decoder to produce explanations directly in the time domain. This methodology builds upon the foundation of L-MAC, Listenable Maps for Audio Classifiers, a method that produces faithful and listenable explanations. We incorporate SepFormer, a popular transformer-based time-domain source separation architecture. We show through a user study that LMAC-TD significantly improves the audio quality of the produced explanations while not sacrificing from faithfulness.
For a prime $p$ congruent to three modulo four, we prove that there exists a smooth curve of genus five in characteristic $p$ that is supersingular. We produce this curve as an unramified double cover of a curve of genus three. We conjecture that the setting of unramified double covers of curves of genus three also produces supersingular curves of genus five when $p$ is congruent to one modulo four, and we computationally verify this conjecture for primes less than $100$. These results can be viewed as a generalization of work of Ekedahl and of Harashita, Kudo, and Senda.
Here we show how to produce a 3D density field with a given set of higher-order correlation functions. Our algorithm enables producing any desired two-point, three-point, and four-point functions, including odd-parity for the latter. We note that this algorithm produces the desired correlations about a set of ``primary'' points, matched to how the spherical-harmonic-based algorithms ENCORE and CADENZA measure them. These ``primary points'' must be used as those around which the correlation functions are measured. We also generalize the algorithm to i) $N$-point correlations with $N>4$, ii) dimensions other than 3, and iii) beyond scalar quantities. This algorithm should find use in verifying analysis pipelines for higher-order statistics in upcoming galaxy redshift surveys such as DESI, Euclid, Roman, and Spherex, as well as intensity mapping. In particular it may be helpful in searches for parity violation in the 4PCF of these samples, for which producing initial conditions for N-body simulations is both costly and highly model-dependent at present, and so alternative methods such as that developed here are desirable