A simple threshold rule is sufficient to explain sophisticated collective decision-making

Decision-making animals can use slow-but-accurate strategies, such as making multiple comparisons, or opt for simpler, faster strategies to find a 'good enough' option. Social animals make collective decisions about many group behaviours including foraging and migration. The key to the collective choice lies with individual behaviour. We present a case study of a collective decision-making process (house-hunting ants, Temnothorax albipennis), in which a previously proposed decision strategy involved both quality-dependent hesitancy and direct comparisons of nests by scouts. An alternative possible decision strategy is that scouting ants use a very simple quality-dependent threshold rule to decide whether to recruit nest-mates to a new site or search for alternatives. We use analytical and simulation modelling to demonstrate that this simple rule is sufficient to explain empirical patterns from three studies of collective decision-making in ants, and can account parsimoniously for apparent comparison by individuals and apparent hesitancy (recruitment latency) effects, when available nests differ strongly in quality. This highlights the need to carefully design experiments to detect individual comparison. We present empirical data strongly suggesting that best-of-n comparison is not used by individual ants, although individual sequential comparisons are not ruled out. However, by using a simple threshold rule, decision-making groups are able to effectively compare options, without relying on any form of direct comparison of alternatives by individuals. This parsimonious mechanism could promote collective rationality in group decision-making.     

Learning the evolutionarily stable strategy

The possibility that animals learn a “developmentally stable strategy” (DSS) (Dawkins, 1980) is an alternative in biological game theory to the idea that evolutionarily stable strategies (ESS) (Maynard Smith, 1972) are genetically determined. A learning rule is defined as a rule which assigns for every possible behaviour the probability of displaying that behaviour at each trial of a game as a function of previous payoffs. This report examines properties of the evolutionarily stable (ES) learning rule, i.e. the rule which, when adopted by a population, is uninvadable by a mutant with a different learning rule. The DSS is defined as the strategy used by individuals with the ES learning rule. With some simplifying assumptions, it is shown that the DSS is the ESS: the ES learning rule is a rule for learning ESSs. This and other properties of the ES learning rule suggested that an approximation to such a rule is the relative payoff sum (RPS) learning rule, which states that the probability of displaying a behaviour is equal to the cumulative payoff for that behaviour relative to the total sum of payoffs for the game. Residual payoffs and a memory factor are incorporated into the RPS learning rule to account for prior expectations of payoff and the decay of memory with time. Both features are adaptive. In simulations of several frequency dependent and frequency independent games using the RPS learning rule, the response of the simulated animals was consistent with the predictions of the ES learning rule. This analysis has shown how ESSs may be achieved by non-genetic means. The RPS learning rule is described in molecular terms utilizing synthesis, storage, and degradation of a substance which elicits the behavioural response. If the RPS learning rule is used by animals, it should be possible to identify within neurons substances whose synthesis is regulated by behavioural stimuli and which initiate alternative behaviours in proportion to their concentrations.     

Adaptive site selection rules and variation in group size of barn swallows: individual decisions predict population patterns

Variation in group size is ubiquitous among socially breeding organisms. An alternative to the traditional examination of average reproductive success in groups of different sizes is to examine individual decision making by determining the cues used for site selection. Once factors used for decision making are known, one can determine whether group-level patterns, such as group size variation, are emergent properties of individual-level decision rules. The advantage of this alternative approach is that it can explain the distribution of group sizes rather than just the occurrence of optimal group sizes. Using barn swallows, I tested, but did not support, the hypothesis that individuals settle at sites based on the previous success of conspecifics (i.e., performance-based conspecific attraction). Instead, I demonstrate that an adaptive site selection decision rule--to breed where it is possible to reuse previously constructed nests--predicts 83% of the variation in the number of breeding pairs at a site. Furthermore, experimental nest removals demonstrated that settlement decisions are also strongly influenced by site familiarity. I discuss the interaction of the cue-based site selection rule with the occurrence of site fidelity and how, more generally, a consideration of individual-level decision rules can improve our understanding of variation in many social behaviors.     

Optimizing Metabolite Production Using Periodic Oscillations

Methods for improving microbial strains for metabolite production remain the subject of constant research. Traditionally, metabolic tuning has been mostly limited to knockouts or overexpression of pathway genes and regulators. In this paper, we establish a new method to control metabolism by inducing optimally tuned time-oscillations in the levels of selected clusters of enzymes, as an alternative strategy to increase the production of a desired metabolite. Using an established kinetic model of the central carbon metabolism of Escherichia coli, we formulate this concept as a dynamic optimization problem over an extended, but finite time horizon. Total production of a metabolite of interest (in this case, phosphoenolpyruvate, PEP) is established as the objective function and time-varying concentrations of the cellular enzymes are used as decision variables. We observe that by varying, in an optimal fashion, levels of key enzymes in time, PEP production increases significantly compared to the unoptimized system. We demonstrate that oscillations can improve metabolic output in experimentally feasible synthetic circuits.     

Callers and satellites in the natterjack toad: evolutionarily stable decision rules

Male natterjack toads, Bufo calamita, sometimes called to attract females and sometimes behaved as silent satellites, attempting to intercept females attracted by callers. Small males with low intensity calls most frequently adopted the satellite tactic and preferred to parasitize the loudest callers. Because body size and call intensity were highly correlated with male mating success, small males with weak calls were at a reproductive disadvantage, and probably achieved greater success by exploiting the displays of larger, louder males than by calling themselves. The behaviour adopted by individual males in a given situation was predicted by a game theory model which assumed that males used an evolutionarily stable decision rule when choosing between alternative mating tactics. However, the model provided close agreement with observed behaviour only when it assumed that males have imperfect information about the behaviour of competitors. It is concluded that male toads adopt a conditional strategy which depends on their attractiveness to females. By switching between mating tactics appropriately, males maximized expected mating success within the constraint of the information available to them. The decision rule of the model qualitatively corresponds to a proximate mechanism that males appeared to use when switching between tactics. The model offers insights into the behaviour of other species in which males display in groups to attract females.     

Classical conditioning,signal detection,and evolution

Strength of classical conditioning is increased either by increasing discriminability of the conditioned stimulus (CS) from the background, or by increasing contingency between conditioned and unconditioned stimili (US).Classical conditioning can be regarded as a decision process in which the subject has to decide whether or not to respond with a conditioned response in the presence or absence of the CS. According to modern evolutionary theories, it might be assumed that this decision process maximizes the trade-off between cost and benefits.By assuming that the decision rule maximizes expected benefit, the empirical relationship between contingency and the strength of classical conditioning is theoretically derived. In addition, when the decision rule is incorporated to a signal detection paradigm, theoretical results describing the relationship between CS discriminability and CS – US contingency with the strength of classical conditioning are in agreement with experimental data.     

Automating parallel implementation of neural learning algorithms

Neural learning algorithms generally involve a number of identical processing units, which are fully or partially connected, and involve an update function, such as a ramp, a sigmoid or a Gaussian function for instance. Some variations also exist, where units can be heterogeneous, or where an alternative update technique is employed, such as a pulse stream generator. Associated with connections are numerical values that must be adjusted using a learning rule, and and dictated by parameters that are learning rule specific, such as momentum, a learning rate, a temperature, amongst others. Usually, neural learning algorithms involve local updates, and a global interaction between units is often discouraged, except in instances where units are fully connected, or involve synchronous updates. In all of these instances, concurrency within a neural algorithm cannot be fully exploited without a suitable implementation strategy. A design scheme is described for translating a neural learning algorithm from inception to implementation on a parallel machine using PVM or MPI libraries, or onto programmable logic such as FPGAs. A designer must first describe the algorithm using a specialised Neural Language, from which a Petri net (PN) model is constructed automatically for verification, and building a performance model. The PN model can be used to study issues such as synchronisation points, resource sharing and concurrency within a learning rule. Specialised constructs are provided to enable a designer to express various aspects of a learning rule, such as the number and connectivity of neural nodes, the interconnection strategies, and information flows required by the learning algorithm. A scheduling and mapping strategy is then used to translate this PN model onto a multiprocessor template. We demonstrate our technique using a Kohonen and backpropagation learning rules, implemented on a loosely coupled workstation cluster, and a dedicated parallel machine, with PVM libraries.     

Confidence scores for prediction models

In medical statistics, many alternative strategies are available for building a prediction model based on training data. Prediction models are routinely compared by means of their prediction performance in independent validation data. If only one data set is available for training and validation, then rival strategies can still be compared based on repeated bootstraps of the same data. Often, however, the overall performance of rival strategies is similar and it is thus difficult to decide for one model. Here, we investigate the variability of the prediction models that results when the same modelling strategy is applied to different training sets. For each modelling strategy we estimate a confidence score based on the same repeated bootstraps. A new decomposition of the expected Brier score is obtained, as well as the estimates of population average confidence scores. The latter can be used to distinguish rival prediction models with similar prediction performances. Furthermore, on the subject level a confidence score may provide useful supplementary information for new patients who want to base a medical decision on predicted risk. The ideas are illustrated and discussed using data from cancer studies, also with high-dimensional predictor space.     

Win-stay-lose-learn promotes cooperation in the spatial prisoner's dilemma game

Holding on to one's strategy is natural and common if the later warrants success and satisfaction. This goes against widespread simulation practices of evolutionary games, where players frequently consider changing their strategy even though their payoffs may be marginally different than those of the other players. Inspired by this observation, we introduce an aspiration-based win-stay-lose-learn strategy updating rule into the spatial prisoner's dilemma game. The rule is simple and intuitive, foreseeing strategy changes only by dissatisfied players, who then attempt to adopt the strategy of one of their nearest neighbors, while the strategies of satisfied players are not subject to change. We find that the proposed win-stay-lose-learn rule promotes the evolution of cooperation, and it does so very robustly and independently of the initial conditions. In fact, we show that even a minute initial fraction of cooperators may be sufficient to eventually secure a highly cooperative final state. In addition to extensive simulation results that support our conclusions, we also present results obtained by means of the pair approximation of the studied game. Our findings continue the success story of related win-stay strategy updating rules, and by doing so reveal new ways of resolving the prisoner's dilemma.     

Sequential Decision Rules for Managing Nematodes with Crop Rotations

A dynamic model of nematode populations under a crop rotation that includes both host and nonhost crops is developed and used to conceptualize the problem of economic control. The steady state of the dynamic system is used to devise an approximately optimal decision policy, which is then applied to cyst nematode (Heterodera schachtii) control in a rotation of sugarbeet with nonhost crops. Long-run economic returns from this approximately optimal decision rule are compared with results from solution of the exact dynamic optimization model. The simple decision rule based on the steady state provides long-run average returns that are similar to the fully optimal solution. For sugarbeet and H. schachtii, the simplified rule can be calculated by maximizing a relatively simple algebraic expression with respect to the number of years in the sequence of nonhost crops. Maximization is easy because only integers are of interest and the number of years in nonhost crops is typically small. Solution of this problem indirectly yields an approximation to the optimal dynamic economic threshold density of nematodes in the soil. The decision rule requires knowledge of annual nematode population change under host and nonhost crops, and the relationship between crop yield and nematode population density.     

Potential importance of vitrification in reproductive medicine

As early as 1985, ice-free cryopreservation of mouse embryos at -196 degrees C by vitrification was reported in an attempted alternative approach to cryostorage. Since then, vitrification techniques have entered more and more the mainstream of animal reproduction as an alternative cryopreservation method to traditional slow-cooling/rapid-thaw protocols. In addition, the last few years have seen a significant resurgence of interest in the potential benefits of vitrification protocols and techniques in human-assisted reproductive technologies. The radical strategy of vitrification results in the total elimination of ice crystal formation, both within the cells being vitrified (intracellular) and in the surrounding solution (extracellular). The protocols for vitrification are very simple. They allow cells and tissue to be placed directly into the cryoprotectant and then plunged directly into liquid nitrogen. To date, however, vitrification as a cryopreservation method has had very little practical impact on human-assisted reproduction, and human preimplantation embryo vitrification is still considered to be largely experimental. Besides the inconsistent survival rates that have been reported, another problem is the wide variety of different carriers and vessels that have been used for vitrification. Second, many different vitrification solutions have been formulated, which has not helped to focus efforts on perfecting a single approach. On the other hand, the reports of successfully completed pregnancies following vitrification at all preimplantation stages is encouraging for further research and clinical implementation. Clearly, however, attention needs to be paid to the inconsistent survival rates following vitrification.     

Shrunken p-values for assessing differential expression with applications to genomic data analysis

In many scientific problems involving high-throughput technology, inference must be made involving several hundreds or thousands of hypotheses. Recent attention has focused on how to address the multiple testing issue; much focus has been devoted toward the use of the false discovery rate. In this article, we consider an alternative estimation procedure titled shrunken p-values for assessing differential expression (SPADE). The estimators are motivated by risk considerations from decision theory and lead to a completely new method for adjustment in the multiple testing problem. In addition, the decision-theoretic framework can be used to derive a decision rule for controlling the number of false positive results. Some theoretical results are outlined. The proposed methodology is illustrated using simulation studies and with application to data from a prostate cancer gene expression profiling study.     

AML-loaded DC generate Th1-type cellular immune responses in vitro

BACKGROUND: The generation of AML-specific T-lymphocyte responses by leukemia-derived DC has been documented by multiple investigators and is being pursued clinically. An obstacle to widespread use of this strategy is that it has not been possible to generate leukemic DC from all patients, and an alternative approach is needed if the majority of leukemia patients are to receive therapeutic vaccination in conjunction with other treatment protocols. METHODS: In the present study, we generated DC from CD14-selected monocytes isolated from healthy donor PBPC and loaded them with a total cell lysate from AML patient blasts. RESULTS: Immature in vitro-derived DC exhibited robust phagocytic activity, and mature DC demonstrated high expression of CD80, CD83, CD86 and the chemokine receptor CCR7, important for DC migration to local lymph nodes. Mature, Ag-loaded DC were used as APC for leukemia-specific cytotoxic T-lymphocyte (CTL) induction and demonstrated cytotoxic activity against leukemic targets. CTL lysis was Ag-specific, with killing of both allogeneic leukemic blasts and autologous DC loaded with allogeneic AML lysate. HLA-matched controls were not lysed in our system. DISCUSSION: These data support further research into the use of this strategy as an alternative approach to leukemia-derived DC vaccination.     

Flight or fight: flexible antipredatory strategies in porcelain crabs

Autotomy, the voluntary shedding of limbs or other body partsin the face of predation, is a highly effective escape mechanismthat has evolved independently in a variety of taxa. Crabs areunusual in that the limb that is typically sacrificed duringautotomy, the anterior clawed cheliped, can also be used toward off attack. During an encounter with a predator, an individualmust thus decide between two mutually exclusive strategies:flight or fight. We used experimental predation encounters withtwo species of porcelain crabs (genus Petrolisthes) to examinethe factors that influence the decision to flee versus fightand to determine the degree to which this decision is context-dependent.We found that autotomy was highly conditional. The characteristicsthat best predicted autotomy—smaller body size or femalegender—also correlated with a lower escape rate by thealternative escape tactic, struggling and pinching the predator.Variation among individuals in the benefit of autotomy (relativeto alternative tactics) appears to drive variation in propensityto autotomize. Porcelain crabs thus demonstrate adaptive flexibility,employing the costly strategy of autotomizing a limb as a lastresort, only when their chance at success by struggling is low.     

Decision making in recurrent neuronal circuits

Decision making has recently emerged as a central theme in neurophysiological studies of cognition, and experimental and computational work has led to the proposal of a cortical circuit mechanism of elemental decision computations. This mechanism depends on slow recurrent synaptic excitation balanced by fast feedback inhibition, which not only instantiates attractor states for forming categorical choices but also long transients for gradually accumulating evidence in favor of or against alternative options. Such a circuit endowed with reward-dependent synaptic plasticity is able to produce adaptive choice behavior. While decision threshold is a core concept for reaction time tasks, it can be dissociated from a general decision rule. Moreover, perceptual decisions and value-based economic choices are described within a unified framework in which probabilistic choices result from irregular neuronal activity as well as iterative interactions of a decision maker with an uncertain environment or other unpredictable decision makers in a social group.     

Assessing the robustness and optimality of alternative decision rules with varying assumptions

Several alternative decision rules have been proposed for how individuals assess and choose options, such as mates and territories. Three of these rules are the threshold rule, where individuals choose the first option that exceeds a preset level of quality, the best-of- n rule, where individuals assess a fixed number of options and then choose the best of those options, and the comparative Bayes rule, where individuals use estimates of options to selectively assess and choose options. It has been previously concluded that the threshold rule produces higher average fitness than the best-of- n rule when assessment costs are not trivial. However, previous comparisons assumed that time and options are infinite, individuals can estimate the distribution of option quality without uncertainty or mistakes, and individuals receive perfect information about the quality of assessed options. I found that the best-of- n rule produces higher average fitness than the threshold rule despite significant assessment costs, when time for choosing an option is limited, when individuals are choosing from a small pool of options, when estimates of the distribution of option quality are error-prone, and when there is uncertainty about the distribution of option quality. I also found that the comparative Bayes rule produces higher average fitness than the threshold and best-of-n rules when time or options are limited and when individuals receive imperfect information about the quality of assessed options. Therefore, the optimality of alternative decision rules depends on more than the size of assessment costs and the previous conclusions of empirical studies that have assumed such need to be re-examined. Copyright 2002 The Association for the Study of Animal Behaviour. Published by Elsevier Science Ltd. All rights reserved.     

BOTTLENOSE DOLPHINS CAN GENERALIZE RULES and DEVELOP ABSTRACT CONCEPTS

Abstract: Generalization of a rule is demonstrated if the rule governs a class of problem, and the subject, after successful experience with a limited number of problems, can apply the governing rule to new problems within that class. We show that the bottlenose dolphin ( Tursiops truncatus ) is capable of such generalization for classes of problems requiting the matching of one of two alternative stimuli to a "sample" stimulus to which the animal had been previously exposed, regardless of the sensory domain used: vision, passive listening, or active echolocation. We also show this generalization capability in a related class of problem requiring a judgment of whether a single "probe" stimulus is the same as, or different from, a stimulus or stimuli previously presented. Further, one dolphin was shown capable of developing a true abstract concept of same/different through its ability to categorize pairs of simultaneously presented objects as identical or not. The suggestion that such generalization ability of dolphins may be in question because of so-called exclusion effects is shown to be not tenable when the whole body of available data is considered.     

Frequency-domain identification of the human controller

System identification techniques applied to experimental human-in-the-loop data provide an objective test of three alternative control-theoretical models of the human control system: non-predictive control, predictive control, and intermittent predictive control. A two-stage approach to the identification of a single-input single-output control system is used: first, the closed-loop frequency response is derived using the periodic property of the experimental data, followed by the fitting of a parametric model. While this approach is well-established for non-predictive and predictive control, it is here used for the first time with intermittent predictive control. This technique is applied to data from experiments with human volunteers who use one of two control strategies, focusing either on position or on velocity, to manually control a virtual, unstable load which requires sustained feedback to maintain position or low velocity. The results show firstly that the non-predictive controller does not fit the data as well as the other two models, and secondly that the predictive and intermittent predictive controllers provide equally good models which cannot be distinguished using this approach. Importantly, the second observation implies that sustained visual manual control is compatible with intermittent control, and that previous results suggesting a continuous control model for the human control system do not rule out intermittent control as an alternative hypothesis. Thirdly, the parameters identified reflect the control strategy adopted by the human controller.     

Optimal Foraging in Hummingbirds: Testing the Marginal Value Theorem

To a hummingbird, clusters of flowers on inflorescences representpatches and provide an ideal situation to test prediction ofoptimal patch-use. The basic question is what decision ruleshould a hummingbird use to decide whether or not to leave aninflorescence? The hypothesis is that hummingbirds will adoptthe decision rule that maximizes their net rate of energy gainwhile foraging. This hypothesis leads to an analogue of Charnov'smarginal value theorem which determines an optimal decisionrule. The optimal decision ruleis then used to predict aspectsof the hummingbirds' foraging, and these predictions are comparedwith field data The optimal decision rule is a function of how much informationis used by the hummingbirds. Data indicate that a decision toleave an inflorescence is a function of the number of flowersvisited, the number of flowers available on the inflorescence,and the amount of nectar obtained at the last flower. The optimaldecision rule was calculated assuming no additional informationis used.     

Decision learning about production control as machines break down in a flexible manufacturing system

During manufacturing, there are many situations that can affect production performance. Such situations include machine breakdowns, rush orders, order changes, and order delays. When such issues occur, one has to make decisions to try to maintain production efficiency. Human decisions tend to be too late and incomplete in such contingencies. Thus a system that can make better decisions in time to maintain production performance is needed. To achieve this objective, the intelligent decision system described in this paper integrates artificial intelligence, an optimization technique, and simulation to solve such problems. The decision-making logic of the intelligent decision system is described by event graphs. It imitates the manner of human thinking. Self-learning of the decision-making process is used to strengthen the decision quality. In this study, a method of rule induction is applied to build up the self-learning system. There are two subsystems included in this system. One is rule generation and the other is knowledge management. A case for machine breakdowns is presented and discussed. A series of tests designed to validate the self-learning system are presented. These demonstrate that a rule induction method is suitable for constructing the self-learning.