Thirteenth Annual Summer Interdisciplinary Conference

Authors, Titles, Abstracts

Listing by speaker

SpeakerAllen, Colin
Author 1Allen, Colin
Indiana University
TitleBiographically plausible data sets and big data in the computational humanities
AbstractThe ability to model words and their relationships by computational means is central to much of the digital humanities. But the methods developed thus far are inadequate if we wish to capture the ways in which experts bring different sorts of expertise in words to bear on the problems of finding meaning in text. For the digital humanities to advance it is necessary to go beyond fascination with larger and larger corpora, by also understanding and modeling the human-scale expertise of readers and authors, as shaped by their exposure to limited "biographically plausible" subsets of much larger digital collections. In this talk, I will describe systems that the InPhO group at Indiana University have built to allow relationships among words to be modeled at multiple scales. Our methods support various analyses, from high-level overviews to detailed digging into specific, historically important arguments. I will introduce the goal of using these methods to provide cognitive models of the reading decisions of domain experts.

SpeakerBickhard, Mark
Author 1Bickhard, Mark
Lehigh University
TitleThe Functional Dynamics of the Central Nervous System
AbstractI will outline a model of the functional micro-dynamics of the brain, and say a few words about how this micro-level model can be integrated into a macro-level functional model. The general model addresses emergent phenomena such as cognition, emotions, and consciousness. The micro-level brain model makes sense of multiple phenomena that are known but are at best anomalous with respect to most standard models of brain function: It is simply false that neurons all function as integrate-and-fire threshold switches, and it is simply false that neurons are the only functional kind of cell in the brain. The model to be outlined naturally integrates such phenomena as: silent neurons that never fire; neurons with non-zero baseline rates of oscillation; volume transmitters and neuromodulators; the multifarious functionality of astrocytes; and so on.

SpeakerCao, Rui
Author 1Nosofsky, Robert
Indiana University
Author 2Cao, Rui
Indiana University
Author 3Cox, Gregory
Indiana University
Author 4Shiffrin, Richard
Indiana University
TitleCategorization and Familiarity Processes in Memory Search
AbstractTwo studies explored the impact of learning on short term memory search. Lists of 1 to 16 pictures were followed by a single picture target or foil. In a varied-mapping (VM) condition, targets and foils could switch roles across trials; in a consistent-mapping (CM) condition, targets and foils never switched roles; and in an all-new (AN) condition, on each trial a completely new set of items formed the memory set. In the first study, subjects were random assigned to one of the conditions for one session. Larger memory sets reduced performance, but for targets this decrease in performance was due almost entirely to the lag with which the target was tested. In the CM condition, foil RTs were invariant with set size, whereas target RTs increased slightly with increasing lag. In the second study, four subjects completed over 30 sessions of CM and VM conditions followed by transfer conditions. In the main condition, 20% of the trials used as a target or foil the test probe from the prior trial. In the VM condition, repeating foils yielded dramatic interference, whereas no such interference was observed in the CM condition. This pattern of results provides evidence for the use of a categorization strategy in the CM condition. An exemplar-retrieval model based on familiarity had been proposed for earlier studies in this paradigm. However familiarity should rise for foils that were tested on the prior trial, so a model based solely on exemplar familiarity would predict reduced performance in CM. A new version of the model 'sums' familiarity and category components and provides a unified account for all the above results.

SpeakerCheng, Patricia
Author 1McGillivray, Shannon
Weber State University
Author 2Cheng, Patricia
University of California, Los Angeles
TitleAnalytic versus Empirical Knowledge of the Concept of Causal Invariance
AbstractThe present paper examines whether we humans learn the concept of causal invariance, the independence of causal powers, from our experiences or know it a priori. Causal invariance plays an essential role in the construction of a parsimonious representation of the causal world, but the environment provides no feedback on how the concept is defined. Causal learning without knowing the concept of causal invariance would be like Alice asking the Cheshire Cat for directions without knowing where she wants to go. There must therefore exist analytic knowledge of the concept in all cognitive systems capable of causal learning. The current paper reports an experiment using a blocking paradigm, taking advantage of the two distinct manifestations of causal invariance for continuous and binary outcome variables. In support of the analytic view, human participants showed more blocking of learning when the same outcome variable was perceived as continuous rather than binary, despite identical experiences with the target outcome except for experience conveying its continuous or binary nature.

SpeakerColunga, Eliana
Author 1Colunga, Eliana
University of Colorado Boulder
Author 2Sims, Clare
University of Colorado Boulder
Author 3Schilling, Savannah
TitleModeling the Emergence and Interactions of Early Word Learning
AbstractEarly word learning may be supported by a developmental feedback loop: the words a child learns early on support the generalization of attentional biases, which in turn guides subsequent word learning. In a series of neural network simulations and a longitudinal behavioral study with toddlers in the lab, we explore the interactions between words learned and word learning biases, and argue that it is this interaction that builds the individual developmental trajectories children follow. First, we look at the development of the shape bias for solids and how its emergence is accompanied by an attentional shift in novel noun generalizations for other solidities, in both networks and toddlers. Second, we look at how the emergence of a shape bias for solids is related to a shift in rate of learning for different kinds of words – shape-based or material-based – in networks and toddlers. Third, we look at how these interactions follow different developmental patterns in typically developing children at risk for language disorders, so-called “late talkers”. Finally, we discuss the implications of this approach in increasing our understanding of language disorders, as well as our ability to improve early diagnosis and the design of individualized intervention plans.

SpeakerCox, Gregory
Author 1Cox, Gregory
Indiana University
Author 2Shiffrin, Richard
Indiana University
TitleModeling Information Accumulation in Recognition Memory: Speed-Accuracy Functions and RT Distributions
AbstractCox and Shiffrin (2012) have argued for a dynamic approach to recognition memory which seeks to extend traditional models of memory to explain the time-course of retrieval. According to this approach, when an item is presented for a recognition decision, features of that item gradually accumulate in a memory probe held in short-term memory. This probe is compared in parallel to traces stored in episodic memory, generating a global familiarity signal that changes over time as more and different information is accumulated in the probe. The characteristic time-course of this signal is shown to closely match the speed-accuracy trade-off functions found by Dosher (1984) and to jointly predict accuracy and RT distributions in large-scale recognition studies reported by Rae, Heathcote, Donkin, Averell, and Brown (in press) and Starns, Ratcliff, and McKoon (2012). Because the parameters of our model can be directly identified with memory processes, a dynamic approach can afford deep insights into recognition: For example, we find both quantitative and qualitative evidence that word frequency effects are best attributed to differences in the diagnosticity of the features used to encode words, rather than increased interference from prior episodes for high-frequency words; we also find evidence that speed instructions influence not just participants' decision bounds, but also the fidelity with which the memory probe is constructed. These insights are not readily available from general RT models (e.g., diffusion or LBA) which can fit these data only by incorporating large amounts of variability in "non-decision" processes (e.g., feature sampling/encoding) that are explicitly modeled within our dynamic approach.

SpeakerDeYoe, Edgar
Author 1DeYoe, Edgar
Medical College of Wisconsin
Author 2Puckett, Alexander
Medical College of Wisconsin
Author 3Ma, Yan
Marquette University
TitleAttentional Field Properties from Single Voxels
AbstractThis study employed a new technique to measure the spatial topography of visual attention that combines a unique “attentional drift” fMRI paradigm with a novel computational model based on single voxel responses in human visual cortex. In attention runs (a), subjects covertly attended and tracked a cued target at 4o or 8o eccentricity within a slowly rotating, dartboard-like array of stimulus segments. In sensory runs (b), the same segment was presented in isolation while subjects attended to a fixation target. Voxels having population receptive fields (pRFs) positioned along the trajectory of the target segment were phasically activated when (a) the focus of attention passed over their pRF or (b) when the isolated stimulus segment passed over the pRF. Thus, the duration of fMRI activation was proportional to (a) the width of the attentional focus or (b) the width of the stimulus segment. To compute the widths precisely, we fit the fMRI timecourse with a waveform generated by a model composed of a gaussian (or DOG) attentional field, the stimulus/task sequence, the voxel's estimated pRF, and the hemodynamic response function. Our results demonstrate that the spatial topography of visual attention can be estimated from single voxels within V1, V2, V3, V4, VO, V3AB, IPS, LO and TO and show that the size of the attentional field scales with eccentricity and visual area. Moreover, voxels in multiple visual areas exhibit attention signals indicating a suppressive surround that is distinct from the response profile observed for the sensory pRF.

SpeakerDixon, Peter
Author 1Dixon, Peter
University of Alberta
Author 2Glover, Scott
Royal Holloway University of London
TitleLimitations in the control of individual and joint rhythmic action
AbstractRhythmic behaviour entails synchronizing repeated action with an internal or external signal. As such, there are three possible constraints on rhythmic action: perception of the signal, the action mechanism itself, and the control process that links the two. In the present research, we used a paradigm in which rhythmic finger movements are performed at a gradually increasing rate. The task can be performed either individually (e.g., left and right index fingers moving in synch) or jointly with a partner. We argue that constraints on these mechanisms can be discerned by examining the form of the function relating phase angle error to oscillation rate. The analysis reveals surprising similarities between individual and joint rhythmic action.

SpeakerDosher, Barbara
Author 1Dosher, Barbara
University of California Irvine
TitleObject attention in moderate precision tasks: An elaborated template model account
AbstractEffects of attention have been widely tested in low precision tasks that discriminate quite different stimuli (i.e., horizontal and vertical pattern orientations), but less often in high precision tasks that discriminate similar stimuli (i.e., with very similar angles). Testing higher precision tasks provides insight into attention mechanisms and their effects across the psychometric function. An elaborated perceptual template model (ePTM; Jeon et al, 2009; Liu et al, 2009) identifies the mechanisms of dual-object attention—deficits in reporting one feature each for two objects compared to two features of one object. Dual object conditions report the orientation of one Gabor object and phase of another; single object conditions report orientation and phase of the same object. Object attention effects were examined in a large data set comprising dual object and single object tests, each with a family of psychometric functions in six different levels of external noise for modestly high precision judgments. The data from the 168 stimulus and task conditions are accounted for by narrowing of the attended template compared to the unattended template—resulting in both asymptotic effects of attention in all conditions and substantial effects across the psychometric function in high external noise associated with external noise filtering. The ePTM, composed of two overlapping templates, nonlinearity, internal multiplicative and additive noise, and decision, provides a principled account of patterns of the effects of attention on the psychometric function sometimes associated with response and contrast gain. Collaborators: Zhong-Lin Lu, Songmei Han, & Shiau-Hua Liu. Supported by MH081018.

SpeakerFoster, James
Author 1Foster, James
University of Colorado, Boulder
Author 2Jones, Matt
University of Colorado, Boulder
TitleAnalogical Reinforcement Learning with Two-Stage Memory Retrieval
AbstractWe have recently proposed a model of analogical reinforcement learning, in which analogical comparison provides the RL algorithm with a measure of relational similarity, and RL provides feedback signals that can drive analogical learning. Especially useful analogies produce new schemas that are added to the memory pool. Model performance is considered with respect to a baseline model in which schema induction is not guided by RL feedback signals. This approach relies on combining structure mapping with exemplar-based learning, which is computationally expensive. The current work addresses this challenge by integrating principles from the two-stage MAC/FAC model of analogical retrieval, in which the first stage uses fast feature-vector similarity to efficiently retrieve a set of candidate exemplars, and the second stage uses structural alignment to determine the best analogical matches. This new implementation enables a more psychologically plausible form of analogical inference from the most useful stored exemplars and schemas to novel situations.

SpeakerHanson, Andrew
Author 1Hanson, Andrew
Indiana University
Author 2Tullis, Jonathan
Indiana University
Author 3Goldstone, Rob
Indiana University
TitleThe Bugcatcher
AbstractWe investigate the features of a simple interactive computer game we call The Bugcatcher. The game contains a hidden (simulated) mechanical linkage between the motion of the mouse and the motion of a "flyswatter" on the screen. The issue is to compare the performance in swatting a randomly-placed "bug" on the screen with and without visual knowledge of the mechanical linkage. Some theories of prompting and retention of prompting information would predict that subjects who have seen the linkage performing will have better performance and much better retention of performance quality than those who do not. We investigate a variety of ways of testing this hypothesis, with results that are not always intuitive.

SpeakerHolden, John
Author 1Holden, John
University of Cincinnati
Author 2Pavlov-Garcia, Olivia
University of Cincinnati
TitleAttractor Dynamics Support Cognition
AbstractHistorically, cognitive science relied on representation as a formative assumption. Representations are semantically imbued entities that serve as a basis for symbolic and/or computational operations in support of cognitive activity. Given their foundational status, it is no surprise that representation and symbolic computation saw critiques over the years. Recently, a cadre of cognitive scientists, partly influenced by the mathematics of dynamical systems, proposed theories relying on an alternate formative assumption: Human cognition and action entail self-organization; dynamic, fluctuating embodied states that persist across a range of time scales. The essential thrust of this view is the hypothesis that certain patterns of persistent, organized biological states could obviate the role representations have traditionally played in supporting cognitive theory. Advocates of this approach often implicate attractors, and attractor dynamics as a straightforward conceptual alternative to representation. The goal of this presentation is to plausibly illustrate how dynamical concepts can do much of the theoretical work that was traditionally ceded to static representations in narratives of classically cognitive activities. Sinusoidal manipulations are introduced to temporal estimation performance, simple response time performance, and speeded word naming performance. The entrainment expressed across these three elementary cognitive activities is examined and characterized with measures of attractor strength. The study outcomes suggest that cognitive performance routinely entails coordination across a range of temporal scales. Cognitive performances are also “sticky”, they spontaneously entrain to environmental and functional regularities. Moreover, the nature of the coupling activity expressed in any given cognitive activity is emergent, in that its expression complies with the idiosyncratic details of task demands.

Speakerjones, matt
Author 1jones, matt
university of colorado
TitleA theory of between-trial variability in diffusion models
AbstractDiffusion models of speeded choice provide excellent accounts of accuracy, the distribution of correct response times (RTs), and their dependence on various experimental factors. However, they cannot account for error RTs without additional assumptions that the drift rate and starting point of the evidence process vary across trials. If this across-trial variability is governed by arbitrary probability distributions, the model becomes unfalsifiable (Jones & Dzhafarov, 2014). Here I offer a theory of across-trial variability based on sequential effects from incremental learning. This theory makes precise predictions about how model parameters vary from trial to trial, thus eliminating the problem of excess flexibility. I explore three types of sequential effects (from learning the response base rate, drift rate, and drift criterion), grounded in the classical interpretation of the diffusion model as a sequential likelihood ratio test. The resulting model can reproduce the benchmark crossover effect (fast errors under speed instructions, slow errors under accuracy instructions) that is held as the primary evidence for across-trial variability.

SpeakerKetels, Shaw L.
Author 1Ketels, Shaw L.
University of Colorado Boulder
Author 2Healy, Alice F.
University of Colorado Boulder
Author 3Jones, Matt
University of Colorado Boulder
Author 4Lalchandani, Lakshmi
University of Colorado Boulder
Author 5Martichuski, Diane K.
University of Colorado Boulder
TitleTesting Two Pedagogical Prescriptions in the Use of Classroom Response Systems
AbstractWe manipulated the usage of classroom response systems, or “clickers,” in four statistics classes taught by the same instructor at the University of Colorado Boulder. In the first experiment, we manipulated the schedule of clicker questions in two classes, taught during the Spring 2013 semester. Clicker questions from a given lecture were either all presented at the end of the lecture (delayed), or interspersed throughout the lecture (interleaved). In the second experiment, we manipulated the format of the clicker questions themselves, comparing the standard multiple choice question format (recognition) to a question format demanding recall of information before selecting among response options (recall). In both experiments, conditions were alternated within-subjects for both classes individually, and these patterns of alternation were counterbalanced between the two classes. Performance on midterm and final exam questions was used as the dependent measure for both experiments. In the first experiment, superior test performance was expected for material presented in the interleaved condition. In the second experiment, superior test performance was expected for material presented in the recall condition.

SpeakerKowler, Eileen
Author 1Kowler, Eileen
Rutgers University
Author 2Zhao, Min
Rutgers University
Author 3Hemmer, Pernille
Rutgers University
TitleTraveling in and out of blind alleys: Strategies of eye movement planning during maze solving and other tasks
AbstractVisuo-motor tasks require eye movements to ensure the line of sight reaches critical details at critical points in time. We recorded eye movements of observers during a challenging visuo-motor task, namely, moving a computer mouse through an overhead maze displayed on a screen. Mazes were randomly generated and difficulty varied considerably. Analyses of eye movements showed that strategies fell into two distinct phases: (1) Guidance, in which the eye made saccades to critical points along the path and the mouse caught up; (2) Exploration, in which the mouse remained relatively stationary while the eye searched for the correct path. Although guidance seems to be more costly because it leads to frequent episodes of traveling down blind alleys, the maze-travelers we tested (with some interesting individual exceptions) were reluctant to devote too much time to exploration, even though exploration had demonstrable benefits for learning the maze. We conclude that the choice of strategies governing how to coordinate movements of hand and eye takes into account multiple factors, including perceived or actual costs of using internal resources to remember or plan the path.

SpeakerLewandowsky, Stephan
Author 1Lewandowsky, Stephan
University of Bristol
Author 2Ecker, Ullrich
University of Western Australia
Author 3Oberauer, Klaus
University of Zurich
TitleEvidence for Removal of Information from Working Memory
AbstractMany complex cognitive activities, such as mental arithmetic, rely on an agile and flexible working memory (WM) that can be cleared of no-longer relevant information upon demand. For example, intermediate steps during mental arithmetic need to be forgotten quickly lest they interfere with the final result. We use a modified version of an established updating paradigm to provide evidence for an active removal process that purges no-longer-needed information from WM. We present experiments that furthermore show that this removal process is an item-specific operation and that updating subsets of information held in working memory involves switching between maintenance and updating modes of processing. We reject other candidate explanations for the data, such as passive decay or displacement by selective rehearsal. We conclude that a full explanation of WM control processes must include a provision for removal of information.

SpeakerLittle, Daniel
Author 1Little, Daniel
The University of Melbourne
Author 2Eidels, Ami
The University of Newcastle
Author 3Fific, Mario
Grand Valley State University
Author 4Wang, Tony
The University of Melbourne
TitleHow do information processing systems deal with incongruent information?
AbstractIn this presentation, we analyze how different information processing architectures deal with incongruent information. For instance, a robust finding is that RTs are slower when dealing with incongruent sources of information (e.g., determining whether a whale is a fish or a mammal? A whale has biological properties which make it a mammal but lives in an environment typically populated by fish) than when dealing with congruent sources of information (e.g., determining whether a trout is a mammal or a fish). We argue that the effect of incongruent information depends on the processing architectures and derive new measure of information processing called resiliency by drawing an analogy to the capacity of an information processing system. By varying the salience of the incongruent information source, we show that serial, parallel and coactive information processing architectures predict qualitatively different resilience functions. We analyze four previously published experiments and show that this measure provides convergent evidence about the underlying processing architecture.

SpeakerLu, Zhong-Lin
Author 1Lu, Zhong-Lin
Department of Psychology, The Ohio State University
TitleQuick methods: Baynesian adaptive methods for estimating psychological functions
AbstractAdaptive procedures are developed to reduce the burden of data collection in psychophysics by creating more efficient experimental test designs and methods of estimating either statistics or parameters. In some cases, these adaptive procedures may reduce the amount of testing by as much as 80% to 90%. For example, adaptive methods for estimating properties of psychometric functions improve test efficiency by targeting stimuli to pre-specified regions of the empirical psychometric functions (e.g. threshold region) based on subject responses. Our goal is to develop adaptive methods for the estimation of psychophysically measured functions and surfaces. In this talk, I will de scribe the Bayesian adaptive framework for optimizing psychophysical tests and its application to the development of various quick methods for measuring TvC functions, d' psychometric functions, contrast sensitivity functions, and forgetting functions. I will provide animations, simulations and psychophysical validations of these methods, and discuss challenges and future directions.

SpeakerLudwig, Kirk
Author 1Ludwig, Kirk
Indiana University
TitleIs Distributed Cognition Group Level Cognition?
AbstractGroups can sometimes solve problems more quickly and efficiently than individuals, and can sometimes solve problems which it is unlikely that any of their members acting alone could solve. Moreover, in some circumstances, it appears that groups exhibit cognitive capacities or mechanisms for problem solving which are different in kind from those exhibited by any of their members. In this paper, I will be concerned with recent arguments (Theiner, Goldstone and Allen, 2010 and related papers), inspired in part by considerations that are at play in the extended mind hypothesis, that seek to move from observations of the sort just mentioned to the existence of genuinely group level cognitive processes and capacities—in the sense that they are an emergent phenomenon which is properly attributed at the group level and that they support the attribution of a group mind, at least so far as cognition goes. I argue that once we have disentangled both the various senses we can give to the notions of a cognitive system, a cognitive process, a cognitive state or property, and a cognitive capacity, and have distinguished among the various kinds of cases that have been offered as concrete instances of group level problem solving, memory, and so on, we can see that there is no sound argument to the conclusion that there are genuine group level minds or cognitive processes except insofar as we interpret that in a Pickwickian sense.

SpeakerMacKay, Don
Author 1MacKay, Don
UCLA Psychology Dept.
TitleCreativity, sentence comprehension and the brain: Lessons from amnesic H.M.
AbstractThis talk presents evidence for strong and previously unsuspected links between the hippocampal region (HR) and the ability to accurately comprehend novel phrases and propositions. The evidence comes from 11 studies with H.M., an amnesic with HR damage but virtually no neocortical damage. Tested was the comprehension of metaphors, ambiguous words, ambiguous sentences, and thematic roles (who-did-what-to-whom in sentences), and comprehension when reading sentences aloud and making judgments of grammaticality. The results identified two types of comprehension process: fast comprehension processes without HR involvement for understanding the meanings of familiar words and phrases, and creative comprehension processes with HR involvement for computing the thematic relations between words and phrases and for integrating familiar word meanings with their sentence context to form useful internal representations of novel phrases and propositions. I discuss the similarities between the creative processes for comprehending sentences and creativity in other domains, e.g., problem solving.

SpeakerNarens, Louis
Author 1Narens, Louis
TitleContext and Probability Theory
AbstractA theory is presented about how context and probability are related in experimental psychology. The matter is not easily treated by traditional probability theory, because contexts have non-traditional logical structure. For example, the intersection of contexts may not be a context, and similarly, the disjunction of contexts may not be a context. (Amos Tversky and colleagues and Jerome Busemeyer and colleagues have exploited the latter in a phenomenon known as the ``disjunction effect’’.) This makes the natural probability theory for contextual studies non-boolean. The talk explains such a “natural probability theory” and its logic.

SpeakerRamsey, William
Author 1Ramsey, William
Philosophy Dept., University of Nevada Las Vegas
TitleRethinking our Conception of Cognition
AbstractIn various contexts and for various reasons, writers often define cognitive processes and architectures as those involving representational states and structures. Similarly, cognitive theories are also often delineated as those that invoke representations to explain various cognitive processes. In this talk, I will present some reasons for rejecting this way of distinguishing the cognitive from the non-cognitive. Some of the reasons against defining cognition in representational terms are that doing so needlessly restricts our theorizing, it undermines the empirical status of the representational theory of mind, and it encourages wildly deflationary and explanatorily vacuous conceptions of representation. If there is time, I’ll consider alternative ways we might try to capture what is distinctive about cognition and cognitive theorizing, though I will also suggest the demarcation problem is far less important than many have thought.

SpeakerRand, Kristina
Author 1Rand, Kristina
University of Utah
TitleAttention Requirements of Navigation with Severely Degraded Viewing
AbstractThe ability to navigate through an environment without getting lost is a challenge for many, but may be even more difficult with vision loss. Low vision research has focused primarily on the low-level detectability of mobility hazards. Although important, visual factors are likely not the sole contributor to difficulty navigating. This work evaluated a cognitive factor – the increased demands of keeping oneself safe while walking with degraded vision (mobility monitoring) – affects navigation and spatial learning. In Experiment 1, participants walked short paths while performing an auditory listening task, half with simulated degraded vision and half with normal vision. Auditory task performance was poorer when navigating with simulated degraded vision, suggesting increased cognitive demands with degraded vision. Experiment 2 tested whether the cost could be attributed to the additional attentional resources needed to walk when mobility-monitoring demands are higher. Participants walked one set of paths while guided (low mobility-monitoring demands) or independently (high mobility-monitoring demands). Importantly, access to visual information was equated for guided and unguided trials; participants wore blur goggles for both conditions. Error rates were higher in the unguided condition, suggesting more attention is required to navigate with high demands of risk monitoring. Experiment 3 used a meditational analysis to test whether attention task errors predicted spatial learning. Participants performed an auditory task while learning the location of landmarks, navigating 2 paths guided and 2 paths unguided. Difference scores on the attention task while navigating guided compared to unguided predicted spatial learning outcomes, providing support for the idea that attentional resources mediates the relationship between mobility-monitoring demands and spatial learning.

SpeakerRouder, Jeffrey
Author 1Rouder, Jeffrey
University of Missouri
TitleDiscrete-states in working memory
AbstractI explore whether visual working memory is mediated by a discrete-slot model, in which items are either in memory or not, or by a resource model, in which the memory for an item reflects the share of latent resources devoted to it. One finding concordant with the discrete-slot model is that in a production paradigm, responses are seemingly a mixture of stimulus-driven responses and guesses (Zhang & Luck, 2008, Psy. Sci). Alternative accounts from the resource theorists is that guessing-like behavior comes from responses to distractor items (Bays et al., 2009, J. Vis) or to items that did not receive many resources (van den Berg et al., 2014, Psy Rev). We use a production paradigm with items that vary in an angular displacement dimension. If guessing occurs, it has a different distribution than either responding to distractors or to responding to targets with a low proportion of resources. We show that the guessing signature remains and cannot be due to these resource-based alternatives. We also show a novel finding: On some trials participants remember the angular disparity finely on a continuous scale, and on others they remember it categorically as left or right.

SpeakerSasaki, Yuka
Author 1Sasaki, Yuka
Brown University
TitleEnhanced Spontaneous Oscillations in the Supplementary Motor Area Are Associated with Sleep-Dependent Offline Learning of Finger-Tapping Motor-Sequence Task
AbstractSleep is beneficial for various types of learning and memory, including a finger-tapping motor-sequence task. However, methodological issues hinder clarification of the crucial cortical regions for sleep-dependent consolidation in motor-sequence learning. Here, to investigate the core cortical region for sleep-dependent consolidation of finger-tapping motor-sequence learning, while human subjects were asleep, we measured spontaneous cortical oscillations by magnetoencephalography together with polysomnography, and sourcelocalized the origins of oscillations using individual anatomical brain information from MRI. First, we confirmed that performance of the task at a retest session after sleep significantly increased compared with performance at the training session before sleep. Second, spontaneous  and fast- oscillations significantly increased in the supplementary motor area (SMA) during post-training compared with pretraining sleep, showing significant and high correlation with the performance increase. Third, the increased spontaneous oscillations in the SMA correlated with performance improvement were specific to slow-wave sleep. We also found that correlations of  oscillation between the SMA and the prefrontal and between the SMA and the parietal regions tended to decrease after training. These results suggest that a core brain region for sleep-dependent consolidation of the finger-tapping motor-sequence learning resides in the SMA contralateral to the trained hand and is mediated by spontaneous  and fast- oscillations, especially during slow-wave sleep. The consolidation may arise along with possible reorganization of a larger-scale cortical network that involves the SMA and cortical regions outside the motor regions, including prefrontal and parietal regions.

SpeakerShiffrin, Richard
Author 1Shiffrin, Richard
Indiana University
Author 2Busemeyer, Jerome
Indiana University
Author 3Wang, Joyce
Ohio State
Author 4Solloway, Tyler
Ohio State
Title A quantum account for a mysterious universal finding
AbstractWhen two questions are asked back to back in a national survey the answers often change depending on the order of the questions (half the respondents are asked the questions in each order). This is a form of 'context effect' and could be part of almost any cognitive model. When looking at all surveys over the last ten years that asked two questions back to back, a peculiar regularity seems to hold for all 70 surveys: The change in the probability of saying yes to both questions plus the change in the probability of saying no to both questions adds to zero. This QQ-equality is not required mathematically; in fact there are surveys that should not and do not show this result (when, for example, extra information is inserted between the two questions). It is hard to come up with any cognitive interpretation or constraints that would require the QQ-equality. We present a quantum model of human decision making that predicts QQ-equality to hold universally, regardless of parameterization. This prediction was derived a priori and verified by results obtained by others years earlier. The fact that the results support the prediction should not only lead cognitive scientists to search for alternative models to explain the finding, but also lead them to give the quantum probability theory serious consideration.

SpeakerShiffrin, Richard
Author 1Shiffrin, Richard
Indiana University
TitleMoving past BMS and MDL: Making model evaluation rational
AbstractI present a generalization of Bayesian Model Selection in which models are treated as greatly simplified but useful accounts. The goal is to use the observed data to induce the probabilities that each model provides the best approximation to the true model and the data produced by that true model. Induction and inference are conditional upon some criterion of ‘best approximation’ that must be chosen in accord with one's goals and task. The approach accords with actual scientific practice, allows one to place the emphasis on data rather than models, allows one to incorporate prior knowledge about both data and models (because there is a direct correspondence between the two), and allows one to incorporate qualitative criteria into quantitative model selection. The key insight is based on the idea that the true model produces a distribution of outcomes for a given experiment, that the observed data are a sample from that true distribution, and that Bayesian induction provides posterior probabilities that each model produces a distribution that best approximates that true distribution, conditional on an observed sample from that true distribution, a goodness of fit criterion, and prior knowledge that allows one to assign prior probabilities to both data distributions and models.

SpeakerSloutsky, Vladimir
Author 1Sloutsky, Vladimir
Ohio State University
TitleLanguage and Cognition: A Whofian baby meets non-whorfian reality
AbstractWhat is the role of language in categorization and category learning? The “Whorfian Infant” hypothesis argues that language (specifically, count nouns) supervises category learning by attracting attention to commonalities among category members. Proponents of the hypothesis assume that the mechanisms underlying effects of labels start operating in early infancy. At the same time, there is evidence that the ability to integrate auditory-visual information exhibits protracted developmental time course, with infants and young children often experiencing difficulty or even failure when such integration is required. In particular, early in development, auditory input (including words) attenuates processing of corresponding visual input. This finding seems to undermine the “Whorfian infant” hypothesis and in attempt to offer an alternative I will focus on three issues. First, I review some data pertaining to cross-modal interference. Second, I consider the role of attention in category learning and how labels may affect attention. Third, I present evidence that early category learning is accompanied by distributed rather than selective attention, which undermines the very mechanism by which labels may affect category learning. Finally, I present data suggesting that labels may supervise learning in adults and discuss how this outcome may emerge in the course of development.

SpeakerSperling, George
Author 1Sun, Peng
University of California, Irvine
Author 2Chubb, Charles
University of California, Irvine
Author 3Wright, Charles E. (Ted)
University of California, Irvine
Author 4Sperling, George
University of California, Irvine
TitleVisual Attention Filters for Color
AbstractAn attention filter is a top-down instruction-initiated brain process of feature-based attention that allows selected visual information to pass but attenuates unselected information. In the centroid-judgment paradigm, subjects use a mouse to position a pointer at the centroid--the center of gravity--of a briefly displayed cloud of dots and receive precise feedback. Centroid judgments are efficient statistical summary representations (SSRs). Our centroid judgment paradigm enables the quick measurement of human perceptual attention filters as accurately as photographic color filters. Here’s how: A subset of dots is distinguished by some characteristic, such as a different color, and subjects judge the centroid of only the distinguished subset, e.g., dots of a particular color ignoring the other colors. A simple linear analysis determines the precise weight to the judged centroid of every dot of every color in the display, i.e., the attention filter for that particular attended color in that context. Here we describe attention filters for colors positioned at various points along orthogonal red-green, blue-yellow, and black-white axes. The selectivities (\"quality\") of attention filters along these three axes are remarkably similar and well predicted by the JNDs between the target color and the nearest distracters. However, attention filters that discriminate among different equi-luminant hues along a color circle are much more selective than predicted from the data described above, and this makes ecological sense. We illustrate here a large set of precise, quantitative data that characterizes processes of feature-attention that previously were mostly described only qualitatively.

SpeakerSteyvers, Mark
Author 1Steyvers, Mark
UC Irvine
Author 2Lee, Michael
UC Irvine
Author 3Liu, Emily
UC Irvine
TitleAggregating Human Judgments in Ranking Problems
AbstractWe develop a cognitive modeling approach, motivated by classic theories of knowledge representation, memory, and judgment, for combining people’s rankings of items. The model makes simple assumptions about how individual differences in knowledge as well as recall processes lead to observed ranking data in behavioral tasks. We implement the cognitive model as a Bayesian graphical model, and use computational sampling to infer an aggregate ranking and measures of the individual expertise. We demonstrate that the model performs well in producing an aggregate ranking that is often close to the ground truth and, as in the ‘‘wisdom of the crowd’’ effect, usually performs better than most of individuals. In addition, we show that it is important to take into account memory processes in top-ten ranking tasks where subjects not only have to rank the items but also have to recall the items that need to be ranked.

SpeakerTeodorescu, Kinneret
Author 1Teodorescu, Kinneret
Indiana University
Author 2Erev, Ido
The Technion - Israel Institute of Technology
TitleLearned Helplessness and Learned Prevalence: Exploring the causal relations of perceived controllability, reward prevalence and exploration
AbstractExposure to uncontrollable outcomes was found to trigger learned helplessness: a state in which the agent fails to take advantage of regained control due to lack of exploration. While the implications of this phenomenon have been widely studied, its underlying cause remains undetermined. One can learn not to explore because the environment is uncontrollable, because the average reinforcement is low or because rewards are rare. The current paper presents a simple experimental paradigm that contrasts the predictions of these three contributors and offers a unified psychological mechanism that underlies the empirical phenomena. The results demonstrate that learned helplessness is not correlated with either perceived controllability or the average reward, suggesting that reward prevalence is a better predictor of exploration behavior. A simple computational model where exploration decisions are based on small samples of past experiences captures the empirical phenomena while also providing a cognitive basis for feelings of uncontrollability.

SpeakerTeodorescu, Andrei
Author 1Teodorescu, Andrei
Indiana University
Author 2Moran, Rani
Tel-Aviv University
Author 3Usher, Marius
Tel-Aviv University
TitleAbsolutely Relative or Relatively Absolute - A study of relativity in decision making
AbstractIn sequential sampling decision models, the evidence accumulation process is terminated by a stopping rule which represents the agent’s response caution. The stopping rule can be applied to the absolute level of activations representing the accumulated evidence for each of the response alternatives. Independent Race and LCA models operate under this assumption. On the other hand, the stopping criterion can also be applied to some function of the relation between the absolute activations. For example, in diffusion models the criterion is applied to the difference while in normalization models it is applied to the ratio. By definition, models implementing relative thresholds are invariant to input manipulations that do not affect the relative aspect of the evidence to which the stopping rule is applied (i.e. difference or ratio). Therefore, a manipulation that only affects the absolute input level without altering its relative aspects could discriminate between relative threshold models, which would predict the null effect, and absolute threshold models, which are sensitive to the overall increase in activation. We present an experimental paradigm which allows for the manipulation of absolute inputs while maintaining constant either their ratio (Multiplicative Boost condition - MB) or their difference (Additive Boost condition – AB). The results reveal a surprising sensitivity to the absolute input. While absolute threshold models naturally account for these combined results, relative threshold models require additional assumptions such as a dependency between processing noise and stimulus intensity. Implications for model architectures, model assumptions and different conclusions about the underlying cognitive mechanisms are discussed.

SpeakerVandekerckhove, Joachim
Author 1Vandekerckhove, Joachim
UC Irvine
Author 2Nunez, Michael
Author 3Baribault, Beth
Author 4Srinivasan, Ramesh
TitleLatent variable methods for data fusion
AbstractWe present a latent variable modeling approach to the joint analysis of behavioral and neural data. The approach relies on defining a fully unobserved structure that determines the covariance between cognitive model parameters and neurophysiological measures. Such measures may be task-dependent ones (to focus on experimental effects) or they may be task-independent (resting state variables to focus on individual differences). The models also allow for the inclusion of external covariates to explain individual differences in addition to experimental design variables. We will present an example data set involving EEG measures in a steady state evoked potential task. We will also discuss issues of robustness under misspecification.

SpeakerWatanabe, Takeo
Author 1Watanabe, Takeo
Brown University
TitleRoles of attention and reward in perceptual learning
AbstractPerceptual learning (PL) is defined as long-term performance improvement on a perceptual task as a result of perceptual experience. We first found that PL occurs for task-irrelevant and subthreshold features and that pairing task-irrelevant features with rewards is the key to form task-irrelevant PL (TIPL) (Watanabe, Nanez & Sasaki, Nature, 2001; Watanabe et al, 2002, Nature Neuroscience; Seitz & Watanabe, Nature, 2003; Seitz, Kim & Watanabe, 2009, Neuron; Shibata et al, 2012, Science). These results suggest that PL occurs as a result of interactions between reinforcement and bottom-up stimulus signals (Seitz & Watanabe, 2005, TICS). On the other hand, fMRI study results indicate that lateral prefrontal cortex fails to detect and thus to suppress subthreshold task-irrelevant signals. This leads to the paradoxical effect that a signal that is below, but close to, one’s discrimination threshold ends up being stronger than suprathreshold signals (Tsushima, Ssasaki & Watanabe, 2006, Science). We confirmed this mechanism by showing that task-irrelevant learning occurs only when a presented feature is under and close to the threshold (Tsushima et al,2009, Current Biol). From all of these results, we conclude that attention and reward play important but different roles in PL.

SpeakerWestfall, Holly
Author 1Westfall, Holly
University of South Florida
Author 2Malmberg, Kenneth
University of South Florida
TitleImprovement in Long-term Memory Retention Created by Learning in an Exploratory State
AbstractWe examined how engaging participants in a visual search task while studying a list of to-be-remembered words affected performance in a subsequent memory task. In a recent publication, we reported that the performance of the search task improved free recall, but the results did not extend to yes–no or forced choice recognition memory testing. These results suggested that the visual search task enhanced the encoding of episodic context information, but not the to-be-remembered item information. In a new set of experiments, we extend our findings to two memory tasks that require the participant to form inter-item associations: a cued recall task, in which participants are asked to learn English-Swahili word pairs, and a serial recall task, in which participants are asked to learn a list of words in the same order they were studied. In both experiments, participants first learned stimuli to a criterion of 100% accuracy, and then when tested after a 48-hour delay, were more likely to remember stimuli for which they had to visually search during study than stimuli learned under standard study instructions. The slower rate of forgetting suggests that visual search facilitated storage of episodic context information.