Inference and Search on Graph-Structured Spaces

How do people learn functions on structured spaces? And how do they use this knowledge to guide their search for rewards in situations where the number of options is large? We study human behavior on structures with graph-correlated values and propose a Bayesian model of function learning to describe and predict their behavior. Across two experiments, one assessing function learning and one assessing the search for rewards, we find that our model captures human predictions and sampling behavior better than several alternatives, generates human-like learning curves, and also captures participants’ confidence judgements. Our results extend past models of human function learning and reward learning to more complex, graph-structured domains.

On Senders’s Models of Visual Sampling Behavior

Objective We review the sampling models described in John Senders’s doctoral thesis on “visual sampling processes” via a ready and accessible exposition. Background John Senders left a significant imprint on human factors/ergonomics (HF/E). Here, we focus on one preeminent aspect of his career, namely visual attention. Methods We present, clarify, and expand the models in his thesis through computer simulation and associated visual illustrations. Results One of the key findings of Senders’s work on visual sampling concerns the linear relationship between signal bandwidth and visual sampling rate. The models that are used to describe this relationship are the periodic sampling model (PSM), the random constrained sampling model (RCM), and the conditional sampling model (CSM). A recent replication study that used results from modern eye-tracking equipment showed that Senders’s original findings are manifestly replicable. Conclusions Senders’s insights and findings withstand the test of time and his models continue to be both relevant and useful to the present and promise continued impact in the future. Application The present paper is directed to stimulate a broad spectrum of researchers and practitioners in HF/E and beyond to use these important and insightful models.

Inference and search on graph-structured spaces

How do people learn functions on structured spaces? And how do they use this knowledge to guide their search for rewards in situations where the number of options is large? We study human behavior on structures with graph-correlated values and propose a Bayesian model of function learning to describe and predict their behavior. Across two experiments, one assessing function learning and one assessing the search for rewards, we find that our model captures human predictions and sampling behavior better than several alternatives, generates human-like learning curves, and also captures participants’ confidence judgements. Our results extend past models of human function learning and reward learning to more complex, graph-structured domains.

Visual Priming Influences Olfactomotor Response and Perceptual Experience of Smells

Whereas contextual influences in the visual and auditory domains have been largely documented, little is known about how chemical senses might be affected by our multisensory environment. In the present study, we aimed to better understand how a visual context can affect the perception of a rather pleasant (floral) and a rather unpleasant (damp) odor. To this end, 19 healthy participants performed a series of tasks including odor detection followed by perceptual evaluations of odor intensity, pleasantness, flowery, and damp characters of both odors presented at 2 different concentrations. A visual context (either congruent or incongruent with the odor; or a neutral control context) preceded odor stimulations. Olfactomotor responses as well as response times were recorded during the detection task. Results showed an influence of the visual context on semantic and motor responses to the target odors. First, congruency between context and odor increased the saliency of the olfactory feature of the memory trace, for the pleasant floral odor only (higher perceived flowery note). Clinical applications of this finding for olfactory remediation in dysosmic patients are proposed. Second, the unpleasant odor remained unaffected by visual primes, whatever the condition. In addition, incongruency between context and odor (regardless of odor type) had a disruptive effect on odor sampling behavior, which was interpreted as a protective behavior in response to expectancy violation. Altogether, this second series of effects may serve an adaptive function, especially the avoidance of, or simply vigilance toward, aversive and unpredictable stimuli.

Ship Sanitation and Sanitary Behavior of the Crew Influences to the Presence of Rats on Cargo Ship in the Port of Tanjung Perak Surabaya

Sanitary inspection intended to eliminate environmental risk factors in the ships to break the chain disease transmission in order to maintain and enhance the health status. Based on data from 2015 the Port Health Office class I Surabaya, during the ships inspection there are found 23 rats from 2734 ships. The existence of the rats on the cargo ship is very harmful, which can caused disease and damaged food material in cargo ship. The tendency of rats existence on board as media transmission of the disease, is the reason for researchers to identify the level of sanitation review and behavior of crew that affecting the existence of rats on cargo ship in the Port of Tanjung Perak Surabaya. That research aimed to analyze the level of sanitation and behavior of crew that affects the existence of rats. This study was an observational study with cross sectional study design, sampling for ship sanitation in research using accidental sampling, behavior of crew using the proportional random sampling method. The research variables are behavior of crew and ship sanitation. The primary data were collected by observation, indepth interviews, and questionnaires. Secondary data collected from the Port Health Office Class I Surabaya. Statistical test with multiple regression showed that behavior of crew significant effect on the existence of rat in cargo ship. The results using ship sanitation level variables sig. 0.043 means that the behavior of crew variable significant effect on the presence of rats on a cargo ship, while for the ship sanitation variables sig. 0.0002, means that the variable ship sanitation very significantly affected on the presence of rats on a cargo ship. The conclusion of this study be found influences of the ship sanitation and behavior of crew against the presence of rats in cargo ship, so it needs to be disseminated to the crew of cargo ship in order to increase knowledge about ship sanitation, and the health effects of the presence of rats on a cargo ship.

Similarity and strength of glomerular odor representations define neural metric of sniff-invariant discrimination time

The olfactory environment is first represented by glomerular activity patterns in the olfactory bulb. It remained unclear, how these activity patterns intersect with sampling behavior to account for the time required to discriminate odors. Using different classes of volatile stimuli, we investigated glomerular activity patterns and sniffing behavior during olfactory decision-making. Mice discriminated monomolecular odorants and binary mixtures on a fast time scale and learned to increase their breathing frequency at a fixed latency after trial initiation, independent of odor identity. Relative to the increase in breathing frequency, monomolecular odorants were discriminated within 10-40 ms while binary mixtures required an additional 60-70 ms. Intrinsic imaging of odor-evoked glomerular activity maps in anesthetized and awake mice revealed that the Euclidean distance between glomerular patterns elicited by different odors, a measure of similarity and activation strength, was anti-correlated with discrimination time. Therefore, the similarity of glomerular patterns and their activation strengths, rather than sampling behavior, define the extent of neuronal processing required for odor discrimination, establishing a neural metric to predict olfactory discrimination time.

Inhalation frequency controls reformatting of mitral/tufted cell odor representations in the olfactory bulb

In mammals olfactory sensation depends on inhalation, which controls activation of sensory neurons and temporal patterning of central activity. Odor representations by mitral and tufted (MT) cells, the main output from the olfactory bulb (OB), reflect sensory input as well as excitation and inhibition from OB circuits, which may change as sniff frequency increases. To test the impact of sampling frequency on MT cell odor responses, we obtained whole-cell recordings from MT cells in anesthetized male and female mice while varying inhalation frequency via tracheotomy, allowing comparison of inhalation-linked responses across cells. We characterized frequency effects on MT cell responses during inhalation of air and odorants using inhalation pulses and also ‘playback’ of sniffing recorded from awake mice. Inhalation-linked changes in membrane potential were well-predicted across frequency from linear convolution of 1 Hz responses and, as frequency increased, near-identical temporal responses could emerge from depolarizing, hyperpolarizing or multiphasic MT responses. However, net excitation was not well predicted from 1 Hz responses and varied substantially across MT cells, with some cells increasing and others decreasing in spike rate. As a result, sustained odorant sampling at higher frequencies led to increasing decorrelation of the MT cell population response pattern over time. Bulk activation of sensory inputs by optogenetic stimulation affected MT cells more uniformly across frequency, suggesting that frequency-dependent decorrelation emerges from odor-specific patterns of activity in the OB network. These results suggest that sampling behavior alone can reformat early sensory representations, possibly to optimize sensory perception during repeated sampling.Significance statementOlfactory sensation in mammals depends on inhalation, which increases in frequency during active sampling of olfactory stimuli. We asked how inhalation frequency can shape the neural coding of odor information by recording from projection neurons of the olfactory bulb while artificially varying odor sampling frequency in the anesthetized mouse. We found that sampling an odor at higher frequencies led to diverse changes in net responsiveness, as measured by action potential output, that were not predicted from low-frequency responses. These changes led to a reorganization of the pattern of neural activity evoked by a given odorant that occurred preferentially during sustained, high-frequency inhalation. These results point to a novel mechanism for modulating early sensory representations solely as a function of sampling behavior.

Active learning reveals underlying decision strategies

One key question is whether people rely on frugal heuristics or full-information strategies when making preference decisions. We propose a novel method, model-based active learning, to answer whether people conform more to a rank-based heuristic (Take-The-Best) or a weight-based full-information strategy (logistic regression). Our method eclipses traditional model comparison techniques by using information theory to characterize model predictions for how decision makers should actively sample information. These analyses capture how sampling affects learning and how learning affects decisions on subsequent trials. We develop and test model-based active learning algorithms for both Take-The-Best and logistic regression. Our findings reveal that people largely follow a weight-based learning strategy rather than a rank-based strategy, even in cases where their preference decisions are better predicted by the Take-The-Best heuristic. This finding suggests that people may have more refined knowledge than is revealed by their preference decisions, but which can be revealed by their information sampling behavior. We argue that model-based active learning is an effective and sensitive method for model selection that expands the basis for model comparison.

Generalization and search in risky environments

How do people pursue rewards in risky environments, where some outcomes should be avoided at all costs? We investigate how participant search for spatially correlated rewards in scenarios where one must avoid sampling rewards below a given threshold. This requires not only the balancing of exploration and exploitation, but also reasoning about how to avoid potentially risky areas of the search space. Within risky versions of the spatially correlated multi-armed bandit task, we show that participants’ behavior is aligned well with a Gaussian process function learning algorithm, which chooses points based on a safe optimization routine. Moreover, using leave-one-block-out cross-validation, we find that participants adapt their sampling behavior to the riskiness of the task, although the underlying function learning mechanism remains relatively unchanged. These results show that participants can adapt their search behavior to the adversity of the environment and enrich our understanding of adaptive behavior in the face of risk and uncertainty.