For undergraduates

Ask your own question, collect the data, and answer it yourself.

Our lab studies how the sleeping brain organizes memory and emotion. Undergraduates take part in that work at every level: some analyze public datasets from another country, some help build research tools online, and students at Kobe University join projects in person. Many of our first-author papers began as undergraduate or medical-student projects (see Where training here leads).

Ways to join

Any university · any year

Remote research

Undergraduates outside Kobe usually start online: analyzing public neuroscience datasets (neural imaging, gene expression), contributing to research tools, and meeting the PI by video. You set the pace with us and report your progress in writing.

Kobe University students

Projects in the lab

Kobe University students of any faculty and year can join projects in person. Medical students can also join through the official first- and second-year research courses. Details are on the Japanese page.

Anyone

Self-study courses

Free 15-minute self-study courses that you can use on your own, with no commitment. See below.

Visiting the lab in person from another institution requires arrangements with your home institution; students under 18 also need a parent's or guardian's consent.

What you can learn here

1. Turn a hunch into a testable question

Most of us feel that a good night's sleep helps the next day. We start by turning such a hunch into a question you can test: which measure, related to what, and by how much.

2. Statistics you understand from the inside

Correlation and p-values, standard deviation versus standard error, the t-test, and how many measurements you need. Underneath them is the central limit theorem: averages of noisy measurements settle into a bell-shaped spread, which is why we can ask whether a difference between averages is larger than chance. You check this yourself by simulating and resampling data.

3. AI as a measuring instrument, and AI you build yourself

Sleep scores from consumer devices are produced by AI. Students first test how accurate such scores are, and those who want to go further build and improve sleep-scoring AI of their own. This year, a second-year student in our lab is working on AI for sleep measurement and on developing the AI itself.

4. Present your work in English

Research presentations in the lab are given in English, and you can prepare in your own language first.

Examples from 2026

Sleep stages and two kinds of athletic performance (second-year student)

The starting point was a feeling from club sports: on days after too little sleep, things did not go well. The student came to the lab in the summer, two months before the second-year research placement began, and started by lining up seven days of sleep records against records from club practice. A few weeks later, the student brought the data they had kept collecting to a meeting and, adding data one day at a time, saw for themselves the moment the relationship became statistically significant. When club practice paused, the student switched the topic to strength training on their own initiative and kept recording.

On days after more deep sleep, the student could lift more weight in the first set of strength training (correlation coefficient 0.85).
The error rate in a ball-sport skill was related not to deep sleep but to the amount of REM sleep. Exerting strength and executing a skill were linked to different stages of sleep.
After the midterm presentation, the student chose an intervention, meditation before bed, started it, and brought in a self-made table comparing before and after. Over the same number of nights (eight each), awakenings of five minutes or longer fell from 1.0 to 0.13 per night, and sleep efficiency also rose (both statistically significant).

Before the final presentation, the student checked each choice in turn, such as whether to report the standard deviation or the standard error alongside a mean and which statistical test to use, until they could explain what every number meant in their own words. In the student's own words: “The nights when I couldn't fall asleep are gone; now I fall asleep right away.” This case showed that one's own sleep can be changed by one's own efforts.

Remote analysis of a public dataset (second-year undergraduate, Australia)

Rickey Arvidson, a second-year undergraduate at the University of Sydney, first wrote to us asking for an internship. Instead, from September 2026, Rickey has worked with us remotely: reanalyzing a public single-cell gene expression dataset of the mouse hippocampus (GEO accession GSE104323), meeting the PI online regularly, and reporting progress in writing. Working only from public data, Rickey independently reached the same answer as our own analysis on one of the questions we are working on, and is now extending it under the PI's guidance.

Building a research tool online (third-year neuroscience undergraduate, USA)

Makoh Miyakawa, a Neuroscience major at Colorado College, joined as an online research volunteer in spring 2026 and is now an affiliated collaborator on Project Commonsound, part of our Sound Exposure during Sleep (SES) research. Across time zones, he leads the English version of the project's sound library and its web form, meets the team weekly, manages the code on GitHub, and checks how sounds play on different phones.

15-minute self-study courses (free for anyone)

Topics that look difficult, such as how AI works or why statistical tests work, can be learned 15 minutes at a time, at your own pace, in conversation. These courses were popular with students in our lab, so we have opened them to everyone.

Self-study with a .md file

Each course is a single .md (Markdown) file: a plain-text teaching file that needs no special software. It opens on a computer or a smartphone, and you can give it to any conversational AI (an AI that answers your questions in a chat). With the file in hand, the AI acts as your tutor and works through the course with you, one question at a time.

Choose a course below and download its file for the AI (click the title to read the course on this site).
Give the file to a conversational AI (attach it, or paste in its contents) and ask: “Please start Session 1.”
Do one session a day, about 15 minutes each (up to 10 sessions per course). The next day, ask: “Please start Session 2.”
Ask the AI about anything unclear as you go. When a session involves a program, ask it to “explain it line by line” and write and run the program together.
After the final session, send us your feedback (see below).

Keep in mind: AI answers can be wrong. Check numbers and facts against the course text on this site, a textbook, or the original source. Do not enter personal information, medical information, or other people's data into the AI.

Each course has a Japanese file, and English files are being added one by one. Give the English file to the AI to learn in English.

If you use a course, please send us your feedback. We revise the courses based on what users tell us. Write to the address on our Contact page with the subject “Course feedback: (course title)”, and tell us how many sessions you completed, what was unclear, what worked well, and how long each session took.

Courses (10 sessions each, 15 minutes per session)

Give a talk people remember — Build your talk from its skeleton and present it without reading a script (files for the AI: English .md · Japanese .md)
Read sleep-and-memory papers with confidence — Build the background to read our lab's papers on your own (files for the AI: English .md · Japanese .md)
Know when a difference is real — See for yourself why a difference between averages can be tested (files for the AI: English .md · Japanese .md)
Sound, sleep and memory: how the research unfolded — Trace how sleep-and-memory research developed, and see what is new in our approach (files for the AI: English .md · Japanese .md)
See how AI learns — hands on — Watch AI learn with short programs, then read a sleep-scoring AI (files for the AI: English .md · Japanese .md)
Read which genes are at work in single cells — Use public data to see, cell by cell, which genes are switched on (files for the AI: English .md · Japanese .md)
Turn videos of glowing neurons into brain activity — Measure the activity of individual neurons from videos of a glowing brain (files for the AI: English .md · Japanese .md)

The courses are free for personal study and teaching. Please contact us before distributing modified versions. Content is current as of November 2026 and is not medical advice.

How to start

Email the PI, Masanori Sakaguchi, with your university, year, and what interests you (address on the Contact page).
We will meet online to connect your interest with a question we can work on together.
In the meantime, try one of the self-study courses above.

For graduate and postdoctoral positions, see Positions.