A conceptual interface mod for Stardew Valley, designed to support neurodivergent players by reducing cognitive friction without lowering difficulty.
Role
Timeline
2 months
Client
N/A (personal project)
Build in
Figma
Stardew Valley is often described as a cozy game. Its pixel farm, unhurried seasons, and gentle rhythm make it a place players return to when they want to relax. Yet beneath that calm atmosphere, the game constantly asks players to keep track of information: birthdays, quest deadlines, crop schedules, animal care, and dozens of small details spread across different menus and moments.
For many players, this is simply part of the experience. For players with cognitive accessibility needs, however, the way information is presented and retrieved can create unnecessary cognitive friction.
This case study explores how interface design can better support those cognitive processes. Through a conceptual mod for Stardew Valley, it proposes a set of interface modifications that reduce cognitive load without changing the game's mechanics or intended difficulty.
The goal is to remove barriers of cognitive friction, not to reduce difficulty. Difficulty is what the game asks you to master: managing the farm, prioritizing, accepting the consequences of a badly planned day. Friction is the mental cost of operating the interface while you do that: remembering, searching, rereading, reacting in time. This project works only on the second.
No proposal forces anything on the player; each one is optional and, where it makes sense, adjustable. Anyone who enjoys keeping track of their own deadlines in their head stays free to do so. The principle: give the player autonomy over pace and load, without changing the task.
This proposal is designed for PC and Mac only, because SMAPI, the mod loader Stardew Valley's community builds on, runs where the game's files are accessible. Consoles don't allow that.
On top of these three pillars sits the cognitive science each category calls on: prospective memory (Einstein & McDaniel, 1990; Fuermaier et al., 2013), visual search and selective attention, time pressure and autonomy. Play Without Barriers (Tencent, 2024) supports the field.
This study ran no primary validation test. The barriers are grounded in three sources: the cognitive literature, an analysis of existing mods, and formative research with players, spontaneous accounts from neurodivergent players on Reddit about their relationship with the game. The validation I would run, once the mod is built, is described in "Next steps and validation."
Beyond the three lenses, some proposals lean on concepts from the design literature:
Glanceable design (Matthews, Forlizzi & Rohrbach, 2006), information understood at a glance, with minimal effort.
Visual variables (Bertin, 1967), the visual grammar (colour, shape, opacity) that encodes meaning.
Information architecture / findability (Morville, 2005), structuring a collection so an item can be found again.
Diegesis (Fagerholt & Lorentzon, 2009), whether the UI lives inside or outside the game's fiction.
The community already feels these barriers. The two reference mods, UI Info Suite 2 and To-Dew, add up to more than 2.2 million unique downloads. But they make information visible, not manageable, and they stay in the pull model.
Reveals hidden information (crop growth, birthdays, animal state) via tooltips and the HUD. It proves the demand, but its philosophy is coverage: everything visible, all the time. It answers "what's the state?", not "what needs my attention now?". More visible information isn't less load.
A board for jotting down tasks. But it's reactive: the player still has to notice the deadline and remember to log it, externalizing memory only after the prospective work is already done.
The clearest evidence isn't in the mods — it's on paper. Players keep notebooks, bullet journals, and spreadsheets to hold what the game won't: bundle checklists, gift preferences, fishing conditions, birthday calendars. When people leave the game to manage the game, the workaround itself is the proof — what they rebuild by hand is exactly what the interface should surface, at the moment it matters.
Search "stardew valley checklist" and brace yourself for a world of spreadsheets, printable planners, and handmade notebooks decorated with pixel villagers.
Six systems, six different places, zero reminders. Welcome to the administrative side of cozy farming.
Plan
organize the tasks into a plan: what to do, and when
Remember
hold the plan in mind until is needed
Start
begin acting at the right moment or cue
Do
carry the plan out as intended
Prospective memory runs in four steps. In adults with ADHD, only the first, planning, breaks down. Remembering, starting, and doing stay intact (Fuermaier et al., 2013).
pull
The quest pin
The player pins an active quest to the HUD. The objective and its days remaining stay in view, so the deadline no longer hides inside a menu.
push
Proactive letters
The game sends a letter the day before a deadline, with the context needed to act. The game does it on its own.
Birthday — a gift hint, from someone close to the villager
Last planting day — before the season closes the window
Dated quest — before a timed quest expires
LOAD:
keeping track
MECHANISM:
push (the game sends)
GAG:"Indicate / allow reminder of current objectives during gameplay" (Cognitive · Intermediate)
Most crops die on the first day of the next season, and each one has its own last possible planting day, counted backward from when it needs to harvest. The game shows the growth time in the seed's tooltip, but doing the math and remembering the deadline is left to the player. The letter arrives on the eve of that day, naming the crop and its deadline. The sender is Pierre, who owns the General Store and sells the seeds: he has reason to keep track.
Since some crops share the same last planting day, their letters can pile up. The frequency is configurable: one letter per crop by default (small notes are easier to act on than a dense list), with the option to switch to a weekly digest.
Stardew Valley has 34 villagers with birthdays, each on a fixed day. A gift on a birthday is worth eight times the usual friendship, and if you miss it, that bonus is gone for a year. Each villager has their own taste: a handful of loved gifts, a few liked ones, and plenty they dislike or hate, across hundreds of items. The date is on the calendar at Pierre's shop, but the player has to remember to check it and figure out what to bring. Two things to hold in your head, almost every other day, all year.
Wrong gift. One year to recover. Or five minutes on the wiki. We’ve all been there.
On the eve of the birthday, a letter arrives, one day early so the player has time to buy or craft the gift. It includes a hint at a gift the villager likes (never one they love, that's left as the player's discovery).
The hint changes by year: in the first year, the letter is narrative, from someone close to the birthday person; from the second year on, it's random, from one of four town figures matched to the gift.
Not every forgotten quest is lost for the same reason. Before the letters, it's worth seeing how quests work today:
A quest can arrive by mail or by a board in town. Some have no deadline (Story Quests); others run against the clock (Help Wanted, Special Orders). Either way, it lands in the journal. The journal button pulses when something new comes in, but after that first signal, it goes quiet. To check the objective, the deadline, or how far along you are, the player has to open the journal, find the quest, and read through it every time.
Every quest ends up in the same journal. Good luck remembering to open it.
The proposal is two new letters, one for each kind of forgetting:
1. The Special Orders letter (quest with a deadline). These are community requests posted on the town board: gather materials, slay monsters, catch fish, within a fixed number of days. Miss the day, lose the reward. The player knows the deadline exists, but the journal never speaks up again after the first pulse.
The letter arrives around the last quarter of the timeframe, early enough to act. It reminds what to do, when it's due, and carries a new hint to help get it done. Signed by whoever posted the order.
Every Special Order was mapped by hand: I wrote a letter template for each one, with variables for the sender, the deadline, the task, and a hint matched to the request. The full spreadsheet holds every combination.
2. The Story Quests letter (quest with no deadline). These are plot-driven quests that arrive by mail: a villager asks the player to find and deliver a specific item, tied to the story and the characters' lives. No deadline sounds forgiving until you forget. Nothing prompts the memory, the quest sinks into the routine, and most ask for a seasonal item. Miss the season, wait a full year.
The letter arrives three days after the quest is received, before the window closes. It puts the quest back in view, naming the item and who to deliver it to, written in the villager's voice so the hint reads as conversation, not instruction. I mapped every story quest in the game and wrote each letter individually.
LOAD:
keeping track
MECHANISM:
pull (the player chooses to pin)
GAG:"Indicate / allow reminder of current objectives during gameplay" (Cognitive · Intermediate).
Sure, Willy. Let me file this right next to the other ten things in my head I'm definitely not going to remember.
A new icon, drawn in the game's pixel style to match the existing ones, marks whether a quest is pinned or not. It sits next to Cancel, inside the quest: one click sends the quest to a compact box on the HUD; another unpins it.
Each pinned quest shows the name, the objective, a live counter, and the days left: exactly what the player would open the journal to check. Up to three fit at once (configurable), and the count shows in the header. When a quest is completed, the pin vanishes on its own.
The pin is pull: the player chooses to use it. To compensate, the empty box stays on screen on purpose, with a shortcut to the journal, so the player doesn't have to remember the feature exists.
The flag that marks a pinned quest uses shape and contrast, not color alone, to tell its states apart (GAG: "Ensure no essential information is conveyed by a fixed colour alone," Vision · Basic).
Every option is configurable through GMCM: how many quests to pin, where the box sits, whether to show the counter and days remaining, and the background transparency. The default works out of the box.
Letters are already saved in Collections since version 1.4, but the archive was built as a record, not a retrieval tool. The proposal adds a dedicated Mail tab with an icon per letter type (quest, gift, event, friendship, info, planting) and filters by type and sender, so the letters from Category 1 stay findable after they leave the mailbox.
It's like Where's Waldo, but every Waldo is wearing the same shirt.
LOAD:
MECHANISM:
GAG:"Allow all narrative and instructions to be replayed" (Cognitive · Advanced)
The proposal is a real mail archive: its own tab in the menu, next to Inventory, Skills, and Collections, always a click away. Inside, each letter gets a small icon set by its intent, so you can tell at a glance what each one is, and the list can be sorted or filtered by type or by sender. The reminder letters from Category 1 keep living here too: when a birthday, planting, or quest letter leaves the mailbox, it stays available to be pulled back.
Six icons, each tied to the context of the letter: a quest, a gift, an event, a relationship, a piece of lore, or a farming reminder. The base game's letters and the mod's share the same icons when the context matches.
Explore the prototype
Dry soil differs from watered by a faint shade; unfed and unpetted animals show nothing at all. The player has to scan every tile and enter every building to check. The proposal adds proximity markers (a half-drop for watering, a half-heart for petting) that pulse near the player and fade with distance, plus a count on the HUD for watering and feeding. Both are glanceable, read preattentively, and fully customizable.
Before getting into the category, it's worth pinning down a few game terms that come up throughout:
Tile
The smallest unit of the game's grid. Each crop bed takes one.
Watering
Using the watering can (manual) or sprinklers (automatic).
Without water, the crop doesn't advance that day.
Feeding
One hay per animal per day, or grass if they graze.
Without food, no produce that day.
Petting
Clicking on the animal. Raises mood and, over time, friendship.
Mood
The animal's daily state. Combined with friendship,
it determines whether produce comes out large and
at what quality (silver, gold, iridium).
Friendship
The long-term bond with the animal. It sets the consistency and level of the products.
Friendship
The animal's daily state. Combined with friendship,
it determines whether produce comes out large and
at what quality (silver, gold, iridium).
Barn / Coop
Barn houses larger livestock (cows, goats, sheep, pigs, ostriches);
coop houses poultry and the like (chickens, ducks, rabbits, dinosaurs)
Hay
The animals' food, cut from grass with the scythe and stored in the silo. One hay per animal per day, when they aren't grazing..
Pop-out search
watering tile almost identical to dry
Checked tile by tile. Cost growns with the field.
Ducan & Humphreys (1989)
Reading the farm - hard
watering tile almost identical to dry
Seein in a single glance
Treisman & Gelade (1980)
Selective visual attention, the system this leans on, is impaired in ADHD, both stimulus- and goal-driven (Klein & Mavritsaki, 2025).
Faint cues + many tiles + a strained attention system = the worst case for these players.
If the barrier is a slow, serial search, the fix is its inverse. Glanceable design is information built to be read "with low cognitive effort," in a single glance, without pulling focus from the main task (Matthews, Forlizzi & Rohrbach, 2006). It works through preattentive processing: the brain categorizes basic features (colour, shape, size, motion) in parallel, before conscious attention, in under ~250 ms (Healey & Enns, 2012; Ware, 2020). The feedback gap forces a tile-by-tile hunt; a glanceable signal is read preattentively, in parallel.
Match expectations
form platers already know
Abstract
stripped to the minimum
Distinct
stands out by contrast
Consistent
one visual language
Proximity markers
A water-drop over each dry plant, a half-heart over each unpetted animal. Faint at a distance, full on approach or hover.
Match
Distinct
Abstract
Consistent
HUD Count
A running count in the HUD: "[N] crops need watering", "[N] animals need feeding".
Match
Abstract
Consistent
LOAD:
keeping track
MECHANISM:
push (ambient/peripheral)
GAG:"Ensure no essential information is conveyed by a fixed colour alone" (Vision · Basic)
"Avoid placing essential temporary information outside the player's eye-line" (Vision · Intermediate) "Provide an option to turn off / hide all non-interactive elements" (Cognitive · Advanced).
The first two are Vision: the project's scope is cognitive, but since the load here is a costly visual search, solving it also benefits players with low vision. The third is the cognitive anchor, the control to reduce or turn off the markers.
Perception and cognition aren't separate systems. When the eye can't catch something automatically, the cost falls on attention and working memory, which is exactly the cognitive system this project deals with. That's why the guideline is "Vision" on paper, but cognitive in effect, for these players.
I thought of two markers, one for each task where reading the state fails:
Watering: an unwatered tile can be hard to see, especially in a large field.
Petting: here there's no marker at all. To know which animals still need petting that day, the player has to test them one by one or open the Animals tab in the menu, which shows it but requires stopping the game to go check.
To reduce this visual search effort, contextual markers provide immediate feedback for daily tasks like watering crops and petting farm animals.
The system operates dynamically based on player proximity and task state:
Explore the prototype. Use the arrow keys or click to navigate.
Explore the quest pin settings menu prototype in Figma
LOAD:
keeping track
MECHANISM:
push + pull
GAG:"Ensure no essential information (especially instructions) is conveyed by text alone, reinforce with visuals and/or speech" (Cognitive · Intermediate).
The goal is to make checking the farm’s state quick and effortless. Instead of moving around the farm to work out what still needs attention, the player gets an immediate overview: a count of crops that need watering or animals that need feeding.
To communicate this information without adding another layer of complexity, the counter follows Stardew’s existing visual language. The game already uses HUD notices that combine text and icons, a pattern that also aligns with the GAG guideline on using multiple modes of communication. The proposal applies the same approach here: a simple count paired with an icon, designed to be understood at a glance.
This turns a scattered check into a single, low-effort piece of feedback. The player does not need to remember where to look, interpret a new system, or inspect the farm to build an overview. The information is available when they need it, in a form that already feels familiar within the game.
Watering
A HUD notice shows how many crops still need water. It appears at a moment set by the player (push) and can also be bound to a key for a manual check (pull). The marker (first proposal) shows which plant; the count shows how many are left, the whole farm in one line.
Feeding
Animals eat by grazing outside or from hay in the trough. The count takes the building's population, subtracts each animal already fed, and shows the rest. Unlike watering, there's no need to find each animal: hay goes to one place, so the useful number is how many are left, not which ones.
The in-game clock runs at a fixed pace the player can't change, compressing decisions into a window that shrinks further for players who perceive time less accurately. The proposal is Day Cycle: the player slows time for a chosen stretch of the day (morning, afternoon, or night), gaining more real seconds per in-game minute without changing the day's structure. The task stays the same; only the pace adjusts.
Morning
Afternoon
Night
what you actually do
Turn the task dial and you get:
• freeze the clock entirely
• remove the fail-state
• skip the task altogether
✗ Easy mode
the challenge is gone, the game unbalances.
how fast you’re forced to act
Turn the rate dial and you get:
• slow the clock for a chosen stretch
• the day still runs 6 am – 2 am
• economy and seasons untouched
✓ Pace control
the challenge is intact, the player has autonomy.
The proposal turns the rate dial, never the task dial.
LOAD:
keeping up
MECHANISM:
GAG:"Include an option to adjust the game speed" (Cognitive · Intermediate).
It's also listed under "Motor · Basic". The project's scope is cognitive, but the same speed control can benefit players with motor limitations.
The proposal lets the player slow that pace for a chosen stretch, morning, afternoon, or night: more real seconds per in-game minute, more real time to act, while the in-game day itself stays untouched. It still runs from 6 am to 2 am, the season still turns, the economy doesn't change. It's regulation, not a longer day.
A snail-shaped indicator below the clock shows when a slowed stretch is active; hovering it brings up the text. Icon and text together keep the information from living in a single channel (GAG: "Ensure no essential information is conveyed by text alone, reinforce with visuals," Cognitive · Intermediate).
Explore all the prototypes in Figma
Once all three modules are released, the validation study becomes executable on two fronts: a controlled pilot and the observation of real use.
Controlled pilot. A within-subject study with neurodivergent players, each completing the same pre-built save twice: once without the mod (Condition A) and once with the mod (Condition B), in counterbalanced order. Before either condition, participants establish a baseline by playing with whatever workarounds they already rely on (notebooks, wikis, spreadsheets on a second screen). This baseline is the real point of comparison, because it captures the cost of the cognitive labor the mod aims to eliminate.
The test uses pre-built saves, not free play. Each save places the player at a moment where the core functions are about to trigger: a birthday tomorrow, a quest approaching its deadline, crops on their last planting day. This guarantees that every participant faces the same scenario and that the difference between conditions comes from the mod, not from where each player happened to be in their own farm.
Participants are experienced Stardew Valley players recruited from neurodivergent communities on Reddit, screened with validated instruments (ASRS for ADHD, GAD-7 for anxiety), excluding anyone already using UI Info Suite 2 or time-control mods. The protocol is concurrent think-aloud. The three comparison measures are described in the diagram below: errors and omissions as the objective anchor, NASA-TLX subscales read as a difference between conditions, and behavioral patterns surfaced through think-aloud observation.
Same save, two conditions, one comparison
BASELINE
Participants use their usual workarounds: notebooks, wikis, or spreadsheets.
Everyone completes the same pre-built save twice. Participants are divided in two groups.
GROUP 1
Condition A — no mod → Condition B — with mod
GROUP 2
Condition B — with mod → Condition A — no mod
03 · COMPARE group 1 VS group 2
Errors & omissions: count how many time-sensitive tasks were missed in each condition
NASA-TLX workload: compare the six subscales, with particular attention to Mental Demand ("how much planning and remembering did the session require?") and Frustration ("how stressed or discouraged did you feel?"). A drop in these two, without a corresponding drop in Performance, would indicate reduced friction without reduced challenge.
Think-aloud themes: track observable behaviors: did the player check the HUD counter before leaving the farm? Did they open a reminder card and act on it, or dismiss it? Did they mention feeling "safer" about forgetting something? These patterns reveal whether the mod changes planning behavior or just adds noise.
Observation of real use. With the mod published on Nexus Mods and GitHub, adoption metrics (downloads, endorsements, unique downloads over time) and community feedback complement the pilot with scale. The format favors this: a free mod with no install barrier lets any player download, test, and report back. Bug reports, configuration requests, and usage descriptions in the Nexus comments section become a lightweight, ongoing signal of what works and what needs revision.
The whole project sits on one distinction: the task versus the friction around it. Accessibility removes friction; easy mode removes the task. In principle the line is clear. In practice, every proposal required checking which side it fell on.
The initial scope was wider. Reading neurodivergent players on Reddit surfaced more pain points than the project could responsibly cover. Fishing was one: the minigame demands rapid, repetitive clicking that many players find exhausting, but the barrier is motor, not cognitive, and the game already offers tools to make it easier. The foraging outline was another: wild items can blend into the scenery, but the game already highlights them at close range. Both were cut because they fell outside a cognitive-accessibility argument. Narrowing the scope was as much part of the work as building what stayed.
The opposite risk was just as present: adding too much. A mod that tries to reduce mental load can create new load with a few extra icons. The markers' detection area, the opacity, the letter frequency are values this design proposes but cannot prove. Where "enough signal" ends and clutter begins is something only testing answers. The fallback was configuration: when an element risked becoming noise, the design hands the choice to the player.
One thing I underestimated. I have over a thousand hours in Stardew Valley. Cues I stopped noticing years ago are real obstacles for someone picking up the game now. Recognizing friction in systems I had already memorized took deliberate effort. The Reddit threads helped ground the analysis in experiences beyond my own, but deep familiarity with a system is both an advantage and a blind spot. That is part of why the validation matters.
One thing I hope this project makes clear: cognitive accessibility deserves to be a routine question in game development, the way visual accessibility is slowly becoming. Accessibility goes beyond contrast ratios and colorblind modes. Disabilities are multiple, often invisible, and the barriers they meet are often invisible too, until someone looks for them. And looking means listening: in Stardew, players spell out their barriers on Reddit every day, and the download counts of accessibility mods are a public record of needs the interface doesn't meet. The research is already there, for anyone willing to read it. The players have already said what's in the way; closing that gap is design work waiting to be done.
Stardew Valley is created by Eric Barone (ConcernedApe), who built not only the game but the conditions this project depends on: a game open to modding and a developer who genuinely listens to his community. This is an unofficial concept study, made with respect and 1,491 hours of gratitude. All game screenshots and assets belong to ConcernedApe.
