THINGS object concept and image database. Object concepts and object images are commonly used for research in psychology, neuroscience and computer science. Existing object concept and object image databases are either limited in the number of concepts, biased towards specific categories such as dogs, have low image quality, or are limited in the number of images per concept.

With the THINGS database, we provide a comprehensive list of 1,854 object concepts used in the everyday American English language, with more than 12 high quality object images per concept, totalling to 26,107 images. In addition, we provide a list of empirically-determined higher level categories of those objects, as well as typicality ratings for objects of those categories. The database is freely available for academic purposes.

List of papers comparing brains, behavior, and DNNs. With the advent of deep convolutional neural networks and their excellent performance at image categorization, there has been a great interest in relating them to human behavior and brain activity. Since 2014, the number of scientific articles have been surging. Here is a list of papers that I collected and that have been published on relating brains, behavior, and DNNs. The list is constantly edited by other researchers, so please add missing papers.

Guide to running Mturk studies. Amazon Mechanical Turk (Mturk) is a popular platform for running online studies, allowing to collect data from thousands of participants in just a few hours. Here are some tips and tricks that should make it easier for you to get started with running those experiments, to save money, and to get higher quality data. Feel free to add comments.