Here, we provide two kinds of data: image- and raw trial-level data. The image-level data is split into four files: 1. **SMID_norms.csv**, which contains scores for each image on a range of psychological variables (e.g., means and SDs for valence, arousal, morality and so-on) 2. **SMID_img_properties.csv**, which contains scores for each image on a range of physical image properties (e.g., average red, green and blue values) 3. **SMID_features.csv**, which contains values for each image regarding a set of high-level visual features (e.g., the presence of humans, animals, etc.) 4. **SMID_metadata.csv**, which contains each image's title, author, and the URL from which it was retrieved The folder **AlexNet Descriptions** contains descriptions of SMID images using a computer vision model designed for object recognition (see [here](https://github.com/damiencrone/smid-alexnet-annotation) for code). This set of features may be useful for applications such as developing predictive models or selecting stimuli in a way that confounding visual features. The raw trial-level data is stored in the **Raw data** folder. There is currently only one file in the folder, but more will be added in future as additional normative datasets are collected. These files contain the individual ratings that are used to construct the normative data, and also contain a small set of demographic variables so that users can compute norms for specific demographic groups (e.g., valence scores for males) to suit their needs.