Signal detection theory, detectability and stochastic resonance effects.

A recent analysis of the energy detector model in sensory psychophysics concluded that stochastic resonance does not occur in a measure of signal detectability ( d'), but can occur in a percent-correct measure of performance as an epiphenomenon of nonoptimal criterion placement [Tougaard (2000) Biol Cybern 83: 471-480]. When generalized to signal detection in sensory systems in general, this conclusion is a serious challenge to the idea that stochastic resonance could play a significant role in sensory processing in humans and other animals. It also seems to be inconsistent with recent demonstrations of stochastic resonance in sensory systems of both nonhuman animals and humans using measures of system performance such as signal-to-noise ratio of power spectral densities and percent-correct detections in a two-interval forced-choice paradigm, both closely related to d'. In this paper we address this apparent dilemma by discussing several models of how stochastic resonance can arise in signal detection systems, including especially those that implement a "soft threshold" at the input transform stage. One example involves redefining d' for energy increments in terms of parameters of the spike-count distribution of FitzHugh-Nagumo neurons. Another involves a Poisson spike generator that receives an exponentially transformed noisy periodic signal. In this case it can be shown that the signal-to-noise ratio of the power spectral density at the signal frequency, which exhibits stochastic resonance, is proportional to d'. Finally, a variant of d' is shown to exhibit stochastic resonance when calculated directly from the distributions of power spectral densities at the signal frequency resulting from transformation of noise alone and a noisy signal by a sufficiently steep nonlinear response function. All of these examples, and others from the literature, imply that stochastic resonance is more than an epiphenomenon, although significant limitations to the extent to which adding noise can aid detection do exist.

In this paper, (supra-threshold) stochastic resonance phenomenon of noise-enhanced information transmission is studied in detail through the numerical calculation and the computer simulation in a non-linear multilevel threshold neural networks system, which is affected by both additive noise and multiplicative noise, then the mutual information is used to characterize the phenomenon. The mutual information as a function of additive noise intensity or multiplicative noise intensity brings on convex changes under a suitable system threshold and a fixed multiplicative noise intensity or additive noise intensity, which shows that the (supra-threshold) stochastic resonance phenomenon occurs. The increases in the number of the system threshold units can enhance the effectiveness of information transmission; the increase of the system threshold can increase the signal components that are under the threshold, and thus the supra-threshold stochastic resonance takes place more easily. In addition, by changing the additive noise intensity the supra-threshold stochastic resonance occurs more easily than by changing the multiplicative noise intensity. The above results show that both the existence of the supra-threshold stochastic resonance and the effectiveness of noise-improved the signal transmission are closely related to multiplicative or additive noise intensity, the number of threshold units, and the system threshold level.

The effect of stochastic resonance is investigated. The principles of the effect are considered on the basis of a Brownian particle moving in a system with a symmetric bistable potential. Different from the widely existing noise elimination methods, stochastic resonance enables a new method to utilize the energy of noises, nonlinear system, and signal frequency to reach certain weak signal enhancement effects that noise elimination cannot perform well. The presence of noise at the input of nonlinear systems possessing the effect of the stochastic resonance allows to stand out a weak signal from an additive mixture with white Gaussian noise. The equation and structural scheme of stochastic resonance are considered The conditions for the occurrence of stochastic resonance are formulated and the determining role of noise variance in the realization of this effect is shown. For the first time, a numerical calculation of standing out of harmonic oscillation from an additive mixture with white Gaussian noise based on the stochastic resonance effect is carried out. The output signal-to-noise ratio dependence of the stochastic resonator on the intensity of input noise and the input signal frequency is investigated. The frequency dependencies of the output signal are investigated and the components of this signal are analysed. It is shown that the stochastic resonator acts as a low-pass filter and reduces the output noise level. Numerical modelling of the stochastic resonance effect also shows that the output signal of the stochastic resonator contains odd harmonics of the input oscillation, which is characteristic of nonlinear systems. It is shown that the stochastic resonance effect can be widely used in radio engineering and telecommunications to amplify the useful signal.

It is well known that when signal and noise sources occupy different spacial positions, detection of the signal is influenced by the orientation of the observer. This paper presents a study of these and related factors as they affect the detectability of single frequency signals using a binaural transmission system. An adjustable-rate pulse-tone method of signal presentation was employed. The two microphones were supported in free space both with and without an artificial head. For the former, various “heads” were used to provide “interaural” separations of 2.5–13 in., while for the latter, where phase differences alone are present, data were taken for linear and rotational separations of the microphones. The results indicate that over-all detectability is a function of the combined binaural phase relationships of both signal and noise and of the monaural intensity relationship of the two signals at the ear having the higher signal-to-noise ratio. In agreement with other studies, the importance of phase differences was found limited to frequencies below about 2000 cps while the importance of pressure-level differences increased as a function of frequency above about 250 cps.

Stochastic resonance in noisy spiking retinal and sensory neuron models

This note analyzes several recent experimental and theoretical reports on vigilance that use the theory of signal detectability (TSD). The psychological interpretation of the TSD measure of criterion, β, as an index of conservativeness during a vigil does not appear to be valid. As computed, β is probably an artifact due to pooling observations made under different conditions of attentiveness during a long vigil. The basic problems of vigilance research remain: to determine the condiitons that affect attentiveness in signal detection tasks.

Drug-associated anaphylaxis is a common condition with significant risks if not promptly addressed. Yet, systematic research on the distribution of associated drugs and risk comparison across drug classes is limited. This study aims to identify frequently reported drugs and evaluate the strength of their signal detections with drug-associated anaphylaxis. This study employed a global pharmacovigilance database to identify reports of drug-associated anaphylaxis. Reports classified as anaphylaxis were analysed using the drug record number used in global pharmacovigilance database, leading to the identification of 15 frequently associated drugs. A disproportionality analysis was conducted to estimate signal detections between these selected drugs and anaphylaxis, utilising two metrics: the information component (IC) with a threshold of IC0.25 and the reporting odds ratio (ROR) with 95% confidence intervals (CI). To account for the acute onset of anaphylaxis, a sensitivity analysis focused on reports with a time to onset of less than a day. We identified 15 drugs frequently associated with anaphylaxis, with diclofenac recording the highest number of reports at 34,413. The drug indicating the strongest signal detection with anaphylaxis was cefuroxime (ROR, 40.89 [95% CI, 40.18-41.61]; IC, 5.14 [IC0.25, 5.11]), followed by levofloxacin, ibuprofen, COVID-19 vaccine, ceftriaxone, lidocaine, omalizumab, cefuroxime, benzylpenicillin, clindamycin, amoxicillin/clavulanate, cefazolin, ciprofloxacin, metronidazole, and paclitaxel. Sensitivity analysis indicated that the signal detection between the COVID-19 vaccine and anaphylaxis was stronger than in the primary analysis (ROR, 2.49 [95% CI, 2.45-2.53]; IC, 1.23 [IC0.25, 1.21]). While most drugs reported that the majority of drug-associated anaphylaxis reports occurred within 2.5 h, omalizumab was often associated with reactions occurring after 24 h. All drugs frequently reported in association with anaphylaxis indicated a significant signal detection, but the strength of these signal detections did not align with the number of reports. Time-to-onset analysis showed distinct patterns for certain drugs, suggesting different mechanisms of anaphylaxis. Due to the limitations of spontaneous reporting databases with disproportionality analysis, our findings do not permit for causal inference.

Signal detection theory addresses the challenge of successfully identifying informative signals in noisy information contexts, allowing optimal behavioural decisions in diverse ecological contexts. The optimal acceptance threshold hypothesis proposed by Reeve (Reeve 1989 Am. Nat.133, 407-435. (doi:10.1086/284926)) is an elegant theoretical model to predict the flexibility of acceptance thresholds for conspecific discrimination. This model has provided a robust framework used to explore recognition systems in a broad range of contexts such as animal communication, nest-mate discrimination or anti-parasitic host responses. In this review, we discuss key concepts related to the optimal acceptance threshold hypothesis applied to egg rejection decisions in avian brood parasite-host interactions. We explore those factors determining signal detectability in parasitized nests and how hosts adjust their rejection decisions to both the risk of parasitism and the potential costs associated with egg rejection. In addition, we discuss recent results that challenge some traditional assumptions of the optimal acceptance threshold hypothesis and provide a novel perspective to explore rejection decisions, such as the existence of single-threshold decision rules or acceptance decisions. An integrative view combining current evidence with traditional theory is needed to further advance the comprehension of optimal acceptance thresholds. This article is part of the theme issue 'Signal detection theory in recognition systems: from evolving models to experimental tests'.

The quality of speech signal needs to be enhanced for applications like teleconferencing, hearing aids, speech recognition etc. Basically, speech-enhancement techniques are responsible for the removal of additive noise, to recover the noise free speech signal. This paper deals with an effective method, stochastic resonance (SR), in which the weak speech signal can be improved by increasing its signal-to-noise ratio (SNR) through addition of white Gaussian noise. Noise, though considered obnoxious in signal processing, can also be used constructively. Pertinence of stochastic resonance in signal processing is aimed at examining the detection of a weak speech signal buried in strong noise. Realizing the effect of stochastic resonance depends on the effective nonlinear system which is mathematically optimized through fourth-order Runge–Kutta algorithm and the system parameters were adjusted adaptively to examine the stochastic resonance (SR) effect on output signal-to-noise ratio and peak of the output signal spectrum.

The asymmetric bistable system with time delays in the feedback force and random force under multiplicative and additive Gaussian noise is studied. Using the small time delay approximation approach and time-delayed Fokker–Planck equations (FPE), the signal-to-noise ratio (SNR) of the proposed stochastic system is obtained. The stochastic resonance (SR) phenomena influenced by parameters — including system parameters [Formula: see text], [Formula: see text], asymmetry parameter [Formula: see text], time delay [Formula: see text], strength [Formula: see text] of the time-delayed feedback, noise intensities [Formula: see text] and [Formula: see text] of multiplicative and additive noise, and correlation strength [Formula: see text] between two noises, are also analyzed by numerical simulations. Results demonstrate that the SR performance of the asymmetric bistable system is superior to one symmetric bistable system. Besides, both time delay and strength of time-delayed feedback could enhance the SR to some extent. Then, the asymmetric time-delayed bistable SR (ATDBSR) method is used to the bearing fault diagnosis. The engineering applications of the ATDBSR method are realized and the value of the method is verified by effective experimental results.

Effects of multiscale noise tuning on stochastic resonance for weak signal detection

In this paper, the phenomenon which is known as stochastic resonance is considered to the signal processing application. Signal standing out with the presence of noise is considered to be one of the basic problems of telecommunication and radioengineering. The stochastic resonance effect is shown to provide significant improvement of some characteristics of the information signal, such as power gain, and noise dispersion at the system output at a certain optimal noise level. In the present article, Minimum Shift Keying signal mixed with Gaussian white noise has been studied using stochastic resonance effect. Noise coefficient, which is one of the quantitative characteristics of noise immunity is calculated and investigated.

In this paper the stochastic resonance (SR) is studied in an overdamped linear system driven by multiplicative noise and additive quadratic noise. The exact expressions are obtained for the first two moments and the correlation function by using linear response and the properties of the dichotomous noise. SR phenomenon exhibits in the linear system. There are three different forms of SR: the bona fide SR, the conventional SR and SR in the broad sense. Moreover, the effect of the asymmetry of the multiplicative noise on the signal-to-noise ratio (SNR) is different from that of the additive noise and the effect of multiplicative noise and additive noise on SNR is different.

Short-term adjustments of signal characteristics allow animals to maintain reliable communication in noise. Noise-dependent vocal plasticity often involves simultaneous changes in multiple parameters. Here, we quantified for the first time the relative contributions of signal amplitude, duration, and redundancy for improving signal detectability in noise. To this end, we used a combination of behavioural experiments on pale spear-nosed bats (Phyllostomus discolor) and signal detection models. In response to increasing noise levels, all bats raised the amplitude of their echolocation calls by 1.8–7.9 dB (the Lombard effect). Bats also increased signal duration by 13%–85%, corresponding to an increase in detectability of 1.0–5.3 dB. Finally, in some noise conditions, bats increased signal redundancy by producing more call groups. Assuming optimal cognitive integration, this could result in a further detectability improvement by up to 4 dB. Our data show that while the main improvement in signal detectability was due to the Lombard effect, increasing signal duration and redundancy can also contribute markedly to improving signal detectability. Overall, our findings demonstrate that the observed adjustments of signal parameters in noise are matched to how these parameters are processed in the receiver’s sensory system, thereby facilitating signal transmission in fluctuating environments.

In this paper, mean first-passage time (MFPT) and stochastic resonance (SR) are investigated in an asymmetric bistable system driven by multiplicative non-Gaussian noise and additive Gaussian white noise. Using the path integral approach and two-state theory, the expression of MFPT and the signal-to-noise ratio (SNR) are derived. The results show that the influences of the asymmetric coefficient on the MFPTs in two opposite directions are entirely different. SNR is a non-monotonic function of the additive noise intensity and asymmetric coefficient, therefore, an SR is found in this system. Whereas SNR is a monotonic function of the multiplicative noise intensity and no SR appears. This demonstrates that the effect of the multiplicative noise intensity on SNR is different from that of the additive noise intensity in the system.