Skip navigation and jump directly to page content
Indiana University Bloomington
Choose site to be searched
Type search terms

Cognitive Lunch

The Cognitive Lunch talks will be on Wednesdays from 12:10 pm - 1:25 pm in the Psychology conference room (PY 128) located behind the main office.

Spring 2015 Cognitive Lunch

Abstract

Jan 28, 2015: Richard Shiffrin
Psychology Rm. 128, 12:10pm
Title: Bayes Factors, Relations to Minimum Description Length, and Overlapping Model Classes
Abstract: Model classes are collections of model instances, usually characterized by parameters with values to be specified (each instance has specific values assigned to all parameters). If we do not compare models that overlap or share instances then model comparison is remarkably simple and coherent, both when using Bayesian Model Selection (BMS) and Minimum Description Length as implemented by Normalized Maximum Likelihood (NML): In BMS we can use the simplest form of Bayes theorem to calculate the posterior probability of each instance, and characterize each class by the sum of such posterior probabilities for its instances. An analogous characterization can be stated for a version of NML, termed NML*. To highlight similarities and difference between BMS and NML we show that the Bayes Factor can be stated as a ratio of the following two measures: For each model class take the average joint probability of instance and observed data across instances, normalized by a sum of like quantities for all data outcomes. NML* can be characterized as a ratio of two NML* scores: For each model class take the maximum joint probability of instance and observed data across instances, normalized by a sum of like quantities for all data outcomes. We finally argue that model comparison generally, including BMS and MDL methods, should not be carried out for classes that share instances, and suggest instead that shared instances in the smaller class be deleted from the larger class prior to comparison. We illustrate the implications with a simple example in which we infer the probability of success, p, based on N trials: One model class assumes p to lie anywhere in [0,1] and the other assumes p to lie only in [0,z].

Feb 4, 2015: Paulo Carvalho
Psychology Rm. 128, 12:10pm
Title: Different study sequences create different attentional biases during category learning
Abstract: We often learn new concepts by studying different examples. Because not all information can be obtained simultaneously, it becomes important to understand how the sequence in which examples are studied can change learning. However, previous evidence on this question has been mixed, with similar study sequences sometimes benefiting and sometimes hindering learning. In this talk I will propose that different study sequences can change what is learned by building different attentional biases on the information presented. Consequently, whether a particular sequence of study helps learners will be the result of whether these attentional biases align with the challenges of the learning task. Using laboratory tasks with novel abstract categories and different study sequences I will present evidence that situations that include frequent category alternation (i.e., interleaved study) or infrequent category alternation (i.e., blocked study) can have an impact on how well people learn. This set of studies demonstrates that frequent alternation of categories during study biases attention towards differences between the categories, benefiting learning situations in which it is hard to discriminate between concepts or when the category assignment is unknown. Conversely, blocked study biases attention towards similarities within each category, benefiting learning situations in which these are hard to detect or in situations that require an isolated representation of the category. Finally, I will discuss the implications of this research for efforts to identify optimal learning sequences across different contexts as well as open questions for future research.

Feb 11, 2015: Ann Bunger
Psychology Rm. 128, 12:10pm
Title: Conceptual Structures Across Languages Influence Language Production of Both Adults and Children
Abstract: When children talk about things happening in the world around them, they have to make choices about what information to convey and how to encode that information in language. These choices are guided not only by how much children understand about the world, but also by developmental constraints and crosslinguistic biases in the way that conceptual structures are mapped to language. In this talk, I will describe two eyetracking experiments that investigate the effect that these factors have on the way that adult and preschool-aged speakers of English and Greek view and describe motion events. The data reveal striking similarities in the conceptual representation of motion events across age and language groups. In addition, they provide a window onto differences in the process of language production in each group by providing a moment-by-moment record of the way that speakers consider possible referents for linguistic encoding.

Feb 18, 2015: Ingrid Van de Leemput
Psychology Rm. 128, 12:10pm
Title: Early warning signals of critical transtions in mood and society?
Abstract: As complex systems, such as the climate or ecosystems, approach a tipping point, their dynamics tend to become dominated by a phenomenon known as critical slowing down. Using time series of autorecorded mood, we have shown that indicators of slowing down are also predictive of future transitions to and from depression. This supports the view that the mood system may have tipping points where reinforcing feedbacks among a web of symptoms can propagate a person into a disorder. In this new communication era where thoughts and feelings of individuals are shared online through social media, one might be able to identfy indicators of critical slowing down in Twitter timeseries, that could function as early warning signals of transtions in mood or society. In this talk, Ingrid will discuss these new ideas, in the hope to receive feedback, and create a discussion.

Mar 4, 2015: Tom Evans
Psychology Rm. 128, 12:10pm
Title: Improving Forecasts of Food Security in Africa through Agricultural Decision-Making Analysis
Abstract: Despite efforts from academic researchers, food security remains an elusive target in many parts of the world, particularly in the face of increasing climate variability. Sub-saharan Africa is especially vulnerable to crop failure because of low annual rainfall and a shortening of the growing season season over the last decade. National governments and international organizations try to forecast where crop failures will occur, but these forecasts are largely driven by physical system dynamics (precipitation, evapotransportation) and have not previously incorporated on-the-ground decision-making dynamics despite the massive influence farm-level decisions have on crop production. A substantial amount of research has addressed whether farmers in developing countries perceive climate change, but the diversity of farming systems globally means the multitude of potential adaptive pathways for farmers present an equally important avenue for decision-making research. This presentation will describe a portfolio of research projects that have focused on decision-making dynamics in small-scale farming systems in Africa. Two components in particular will include description of an agent-based model of smallholder crop production and emerging team-based research on crop forecasting that incorporates farm-level, high spatial resolution agricultural decisions over large spatial extents.

Mar 11, 2015: Melody Dye
Psychology Rm. 128, 12:10pm
Title: Design Constraints on Linguistic Systems: The Case of Common Nouns & Proper Names
Abstract: According to a common framing, language is a vehicle for encoding our thoughts and decoding those of others, or of ‘packing’ and ‘unpacking’ the stuff of thought into linguistic form. In this talk, I will discuss how information theory can offer a reformulation of the traditional ‘code model’ of communication. On this view, meaning does not reside in words or sentences, but in the exchange – and progressive alignment – of speakers with similar codes. Such a perspective casts human languages as systems of social exchange that have evolved both to optimize the flow of information between speakers, and to balance the twin demands of comprehension and production. In support of this approach, I will report on a pair of cross-linguistic projects: one, contrasting the evolution of naming systems in the East and in the West, and the other, comparing the functional role of grammatical gender with that of prenominal adjectives across two Germanic languages. This work suggests a principled means of beginning to piece apart those evolutionary pressures on language that are cognitive and universal, from those that are extrinsic and confined to specific social environments.

Mar 25, 2015: Robert Bowers
Psychology Rm. 128, 12:10pm
Title: Some observations about feeding behaviour in Norway rats in standard conditioning experiments
Abstract: Whether and how to apply theories of causal reasoning to the behaviour of rats in standard food conditioning experiments remain controversial. With a series of experiments involving rats receiving food reward in Skinner Boxes, I attempt to situate a specific analysis of causal reasoning within the independent conceptual framework of behaviour systems. Three questions that arise in the context of theories of causal reasoning are addressed. These regard graph surgery, intervention, and independence of associations and events. Assumptions of theories of causal reasoning are challenged, and behavioural variables beyond those theoretically accessible to standard approaches are shown to be relevant to the behaviour of rats solving causal problems. Behaviour systems is neutral to typical assumptions of theories of rationality, and permits interpretation of a broader range of behaviour, and so provides an apt theoretical context in which to understand how rats deal effectively with cause and effect in causal reasoning experiments.

Apr 8, 2015: Brandi Emerick
Psychology Rm. 128, 12:10pm
Title: Effects of cannabis on object recognition
Abstract: Cannabis is one of the most commonly used drugs in the world; in 2013, 19.8 million people reported using marijuana within the past month in the U.S. alone, according to a national survey by the Substance Abuse and Mental Health Services Administration. Changes in sensory perception are frequently experienced by cannabis users, still little is known about how cannabis affects the visual system. My research uses EEG and frequency tagging to measure brain responses during an object recognition task in which a degraded image presented in noise. Using preliminary data collected from current cannabis users and non-users, I will show how frequency tagging can be used to separate and quantify brain responses to signal and noise during object recognition. These brain responses can then be compared across groups (current, former, and non-users) to better characterize the endocannabinoid system's role in vision.

Apr 22, 2015: Alex Gates
Psychology Rm. 128, 12:10pm
Title: Complex Systems Approaches in Cognitive Science
Abstract: Motivated by attempts to characterize the organization of dynamical processes underlying cognition, this talk explores the interplay between a cognitive system's network structure and the dynamics which unfold. Here, I will introduce two projects which have adapted such a complex system's philosophy. The first explores the relationship between the structure of semantic memory and the dynamics of semantic fluency. Here, several methods for inferring the structure from the dynamics will be explored and characteristics of the dynamical process will be related to inferred community structures. The second project focuses on the structure of ontogenies in a model protocell and thus gives a quantitative method for mapping out the space of possible behaviors.


Previous Cognitive Lunch