Intro

Overview

During the first day, you learned that different models can answer different questions. That means that depending on your question, goals, and hypotheses, you will need to develop different kinds of models. How to best approach this is the goal of today. We will use this as an opportunity to kick off your group projects simultaneously. To do so, we will start walking you through a 10-steps guide of how-to-model. This guide is applicable for both computational modeling and data neuroscience projects, and we will discuss similarities and differences between both project types. So today, we will start with what you now know determines the choice of modeling or data analysis pipeline you will need to make: how to develop a good question and goal, do the literature review, think about what ingredients you need, and what hypotheses you would like to evaluate.

Today’s tutorial focuses on the first 4 steps of how-to-model by demonstrating the thought process based on a simple phenomenon known as the train illusion. We will first introduce the phenomenon and then provide a step-by-step guide on thinking about and developing the 4 first steps of framing your project. To help you, we will roleplay an example thought process, focussing on typical pitfalls that groups often encounter. Groups will then think about their own projects and develop first answers to each step’s questions. Importantly, this is to get you started systematically with your projects; you will have to revisit those steps as your thinking evolves, possibly multiple times. We will also provide similar guidance for the remaining 6 steps of the how-to-model guide that you can work through with your group when you’re reaching that stage of your project. The accompanying answers to each step in our demo project and toy example code for our two different projects (model and data analytics) should help you understand the practical side of the process better.

How to model is rarely, if ever, taught systematically. Our guide is not the only way to approach modeling, but it’s one way to ensure you don’t miss anything important. Going through all the steps also makes publication much easier because you have already explicitly thought about all the elements you will ultimately need to communicate (see Step 10 later for our examples). Personally, I often take shortcuts in this process and then regret it later… mostly because I forgot to do the one most important thing: be precise about the framing of the project, i.e., the four first steps you will walk through today. Importantly this will set you up to develop any kind of model using any of the tools you will learn about during the remainder of NMA.

Install and import feedback gadget

Show code cell source Hide code cell source

# @title Install and import feedback gadget

!pip3 install vibecheck datatops --quiet

from vibecheck import DatatopsContentReviewContainer
def content_review(notebook_section: str):
    return DatatopsContentReviewContainer(
        "",  # No text prompt
        notebook_section,
        {
            "url": "https://pmyvdlilci.execute-api.us-east-1.amazonaws.com/klab",
            "name": "neuromatch_cn",
            "user_key": "y1x3mpx5",
        },
    ).render()


feedback_prefix = "W2D1_Intro"

Video

Slides

Submit your feedback

Show code cell source Hide code cell source

# @title Submit your feedback
content_review(f"{feedback_prefix}_Video")