Decoding motor responses from neural spike activity (2023).

• My project in this biomedical engineering class aimed to decode and understand the motor responses of rats in reaction to tilting events using neural spike activity from the motor cortex. The approach involved several key steps, starting with organizing the recorded spike time data and then creating perievent rasters and histograms to visualize the neural activity. I used a Peri-Stimulus Time Histogram (PSTH) classifier to categorize the different motor responses to clockwise and counterclockwise tilts. To enhance classification performance, I integrated Principal Component Analysis (PCA) with the PSTH classifier. This methodological framework was presented in the work of Foffani and Moxon (2004).

Frequency-specific functional couplings of the Frontal Eye Field and Inferior Frontal Junction (2023).

• For my master’s thesis project at the University of Trento (supervised b Dr. Daniel Baldauf), I worked on the dissociative functional roles of the Frontal Eye Field (FEF) and the Inferior Frontal Junction (IFJ) within the lateral prefrontal cortex using resting-state MEG recordings of 55 subjects. I analyzed the functional phase and power coupling of these regions, in addition to their directed influences with the ventral and dorsal visual streams. We found that FEF and IFJ have predominant functional relation with the dorsal and ventral visual streams, respectively. Our findings were later published in the European Journal of Neuroscience!

Object vs Scene Perception (2023).

• During my internship at CIMeC, we aimed to determine whether images containing both objects and scenes belong to the object or scene category. I employed neural networks such as ResNet and AlexNet, initially trained on object classification (ImageNet) or scene classification (Places365) tasks. I fine-tuned these networks using a custom dataset of object and scene images. For evaluation, I utilized a custom testing dataset containing images of objects, scenes, and both conditions. Features were extracted from each layer of the neural networks and assessed using representational similarity analysis (RSA) to identify which model (control, scene, or object) best explained the variance in the feature space.

Naturalistic video reconstruction from fMRI activity using eye-tracking (2022).

• My research at the Donders Institute focused on advancing methods for the naturalistic reconstruction of freely viewed movies. I used the studyforrest dataset, which includes fMRI scans and eye-tracking data from participants watching the Forrest Gump movie. We aimed to move beyond traditional methods, where subjects fixated on the center, towards a more naturalistic video reconstruction approach. The process involved preprocessing the movie frames based on eye gazes, transforming BOLD responses into a tensor in pixel space, determining the receptive field locations and signals in the early visual cortex, and generating target frames for these signals. These frames were then matched to be input into the brain2pix neural network model for video reconstructions.

Interactive learning environment for teaching English to kids (2020).

• For my final project in an Introduction to MATLAB course during my master’s degree, I developed an interactive program using Psychtoolbox designed to teach English to children. The program is divided into two main sections. The first section focuses on learning activities, including alphabet recognition, vocabulary building, and writing practice. The second section provides exercises to reinforce alphabet and vocabulary knowledge, identify and correct mistakes, and track progress by monitoring errors in letters, vocabulary, and individual examples over multiple attempts.

Information flow between ventral and dorsal streams during complex tool use (2020).

• This is a research proposal that I wrote for the final project for Motor Cognition course during my master’s degree at University of Trento. It focuses on the integration of semantic and sensory-motor information across the ventral and dorsal streams, employing magnetoencephalography (MEG) to overcome the temporal limitations of fMRI. The core objective is to reveal the connectivity and information flow within the brain’s Tool Processing Network during various action stages.

The explanatory gap and code-duality: Reconsidering the mind-body dualism (2017).

• A paper I wrote for an independent study during my bachelor’s degree in philosophy at Bogazici University. It introduces the concept of code-duality, suggesting that our inherent subjectivity and cognitive limitations prevent us from fully understanding the interaction between mind and body. By distinguishing between the phenomenal world (our subjective experiences) and the noumenal world (reality beyond our perceptions), and discussing the emergence of life and self-reference, the paper critiques the notion of an ontological divide between mind and body.