Norris, D. | Search bioRxiv

bioRxiv2018

Is reading automatic? Are the ERP correlates of masked priming really lexical?

Humans have an almost unbounded ability to adapt their behaviour to perform different tasks. In the laboratory, this flexibility is sometimes viewed as a nuisance factor that prevents access to the underlying cognitive mechanisms of interest. For example, in order to study \"automatic\" lexical processing, psycholinguists have used masked priming or evoked potentials to measure \"automatic\" lexical processing. However, the pattern of masked priming can be radically altered by changing the task. In lexical decision, priming is observed for words but not for nonwords, yet in a same-different matching task, priming is observed for same responses but not for different responses, regardless of whether the target is a word or a nonword (Norris & Kinoshita, 2008). Here we show that evoked potentials are equally sensitive to the nature of required decision, with the neural activity normally associated with lexical processing being seen for both words and nonwords on same trials, and for neither on different trials. (150)

neuroscience↗

bioRxiv2017

Visual recency bias is explained by a mixture model of short term memory

Human bias towards more recent events is a common and well-studied phenomenon. Recent studies in visual perception have shown that this recency bias persists even when past events contain no information about the future. Reasons for this suboptimal behaviour are not well understood and the internal model that leads people to exhibit recency bias is unknown. Here we use a well-known orientation estimation task to frame the human recency bias in terms of incremental Bayesian inference. We show that the only Bayesian model capable of explaining the recency bias relies on a weighted mixture of past states. Furthermore, we suggest that this mixture model is a consequence of participants failure to infer a model for data in visual short term memory, and reflects the nature of the internal representations used in the task.

animal behavior and cognition↗

bioRxiv2017

Porcupine: a visual pipeline tool for neuroimaging analysis

The field of neuroimaging is rapidly adopting a more reproducible approach to data acquisition and analysis. Data structures and formats are being standardised and data analyses are getting more automated. However, as data analysis becomes more complicated, researchers often have to write longer analysis scripts, spanning different tools across multiple programming languages. This makes it more difficult to share or recreate code, reducing the reproducibility of the analysis. We present a tool, Porcupine, that constructs ones analysis visually and automatically produces analysis code. The graphical representation improves understanding of the performed analysis, while retaining the flexibility of modifying the produced code manually to custom needs. Not only does Porcupine produce the analysis code, it also creates a shareable environment for running the code, in the form of a Docker image. Together, this forms a reproducible way of constructing, visualising and sharing ones analysis. Currently, Porcupine links to Nipype functionalities, which in turn accesses most standard neuroimaging analysis tools. With Porcupine, we bridge the gap between a conceptual and an implementational level of analysis and thus create reproducible and shareable science. We give the researcher a better oversight of their pipeline, both while developing and communicating their work. This will reduce the threshold at which less expert users can generate reusable pipelines. We provide a wide range of examples and documentation, as well as installer files for all platforms on our website: https://timvanmourik.github.io/Porcupine. Porcupine is free, open source, andreleased under the GNU General Public License v3.0.\n\nAuthor SummaryThe neuroimaging community is fervently debating that its reproducibility and transparency should be improved, but it is a challenging problem as to how to accomplish this. We here propose a tool, Porcupine, to aid in this process by more easily creating shareable workflows for analysing neuroimaging data. The conceptual understanding of a pipeline is improved by means of the graphical interface, and it automatically produces the code to perform the analysis and to create a sharing environment. This retains full flexibility to modify the script afterwards but in principle produces readily executable code for an end-to-end analysis. Porcupine currently links to all Nipype functionality, but is designed to be extendable to other workflow packages in neuroimaging and beyond. Porcupine is free and is released under the GNU General Public License.

neuroscience↗