Counselor intelligence brief
A practical guide for high school students who have a serious STEM project idea, need funding and mentorship to execute it, and want to turn an early-stage proposal into a credible application-season artifact.
Updated June 3, 2026. The 2025-26 cycle had already closed and finalists had been announced. Treat date-specific guidance here as a next-cycle watch item until THINK posts fall 2026 instructions.
THINK is a science, research, and innovation program run by MIT undergraduates. Students apply with a project proposal before the work is finished. Finalists then receive mentoring and modest funding to carry the project through the spring.
The college-application value comes from the proposal judgment: the student has found a real problem, read enough background to understand the field, and designed a semester-scale project that can be tested, built, or measured.
Proposal length
10 pages
THINK describes a proposal limit of up to 10 pages before references, cover pages, and figures.
Finalist support
$1,000
Finalists can receive up to $1,000 in project funding, plus weekly mentorship.
2025-26 deadline
Jan. 1
The 2025-26 cycle listed January 1, 2026 as the application deadline. Verify the next cycle in fall 2026.
THINK is worth a December push only if the student clears the basic rules and the project can realistically fit the program's mentorship-and-funding model.
No data leaves your browser.
Students often pitch a full platform, cure, or national deployment. THINK is stronger when the proposal describes one feasible proof of concept with a clear first experiment, build, dataset, or prototype.
The proposal has to feel novel, useful, and executable. A strong THINK idea explains what exists, what remains unsolved, what the student will do, and how progress will be measured.
Build the application around four proof points. Each one should make the project feel easier to execute, not larger.
01
A specific need, constraint, or scientific question with enough background for an MIT undergraduate reader to care.
02
A short map of existing methods, products, papers, or prototypes, with a clear explanation of the remaining gap.
03
The proposed project, written as a buildable or testable plan instead of a broad promise.
04
Budget, materials, data access, weekly execution plan, risks, safety, mentor needs, and success criteria.
June to August
Problem search
Find a problem that can be studied, built, or tested within one semester and about $1,000. Start with the bottleneck, user, dataset, organism, device, or environment.
September
Background read
Read current literature, comparable products, open-source methods, and safety constraints. The student should know what exists and where the gap sits.
October
Project design
Turn the idea into inputs, procedure, validation metric, budget, week-by-week timeline, and a fallback if the first method fails.
November
Draft
Draft once applications open. Keep figures, tables, budget notes, and citations close to the procedure so the plan feels executable.
December
Counselor review
Review novelty, feasibility, and fit. Trim broad ambition, add experimental detail, and make the mentoring need specific.
January
Submit
Submit before the posted deadline. For 2025-26, THINK listed January 1, 2026 as the deadline.
Most weak proposals fail before the technical idea gets a fair read.
The student proposes a platform, cure, national deployment, or full product instead of a semester-scale proof of concept.
The proposal says the idea is new but never names the closest existing work or the exact improvement.
The application has motivation and goals, then skips materials, data, steps, tests, and constraints.
"I want MIT guidance" is too broad. Stronger proposals explain what technical decision a mentor could help resolve.
Use official THINK materials at the moment of submission. The THINK resources PDF endpoint returned a 502 during this build, so confirm the PDF link again before a student starts drafting.