The Mental Math Problem: Why We're Bad at Splitting Bills

Your brain wasn't designed for restaurant math. Here's why splitting bills overwhelms working memory—and why "just split it evenly" becomes the default.

Seven, plus or minus two

In 1956, cognitive psychologist George Miller published one of the most cited papers in psychology. His discovery: human working memory can hold about 7 ± 2 items at any given time.

This limit isn’t about intelligence. It’s architectural. Your brain’s “RAM” has finite capacity. Exceed it, and information drops out.

A typical restaurant split requires tracking:
• Your items (2-4 things)
• Everyone else’s items (12-20 things for 6 people)
• Who shared the appetizers
• The subtotal
• Tax rate (varies by location)
• Tip percentage
• Who paid what already
• Each person’s final total

That’s 20+ items to track—in a system designed for 7. No wonder splitting bills feels overwhelming.

“My problem is that I have been persecuted by an integer. For seven years this number has followed me around, has intruded in my most private data, and has assaulted me from the pages of our most public journals.”
— George A. Miller, “The Magical Number Seven, Plus or Minus Two,” 1956

Source: Miller, Psychological Review, 1956

Cognitive load theory

In 1988, educational psychologist John Sweller formalized what teachers had always known: when information exceeds working memory, learning—and performance—collapses.

He identified three types of cognitive load:

Intrinsic Load

The inherent complexity of the task. Restaurant math has high intrinsic load—percentages, decimals, multiple operations.

Extraneous Load

Distractions and inefficiencies. Noisy restaurant, friends talking, waiter hovering, social pressure to hurry.

Germane Load

Mental effort devoted to learning and building understanding. At a restaurant, none of your effort builds lasting schemas—it’s all one-off calculation.

At a restaurant table, intrinsic and extraneous loads hit simultaneously. The math is complex (intrinsic), the environment is distracting—noise, social pressure, worry about fairness (extraneous). And there’s zero germane benefit: you won’t remember this math tomorrow. Your working memory is completely saturated for a one-time calculation.

The result: You round numbers. You estimate. You “round up a bit to be safe.” Or you give up entirely and say “let’s just split it evenly.” The cognitive load is too high for precision.

The math anxiety factor

For many people, bill-splitting isn’t just cognitively hard—it’s emotionally triggering. Ashcraft and Moore (2009) estimated that roughly 17% of the population qualifies as highly math-anxious—scores one standard deviation above the mean. And up to 93% of US adults report some level of math anxiety (Blazer, 2011).

Math anxiety is real and measurable. Lyons and Beilock (2012) used fMRI to show that anticipating math activates the posterior insula—a brain region associated with physical pain. When you’re math-anxious, your working memory shrinks because the anxiety itself consumes cognitive resources (Ashcraft & Kirk, 2001).

17%of adults experience high math anxiety

93%of US adults report some level of math anxiety

2xerror rate on carry problems under high math anxiety vs. low (Ashcraft & Kirk, 2001)

The cruelest irony: people most likely to be shortchanged by equal splits (modest orderers, careful calculators) are often the same people least likely to speak up and do the math publicly.

Math anxiety affects financial decisions. Choe, Jenifer, Rozek, Berman, and Beilock (2019) at the University of Chicago found that math-anxious people consistently avoid harder math problems even when they pay higher rewards. They’d rather lose money than do the math. This is why “just split it evenly” wins so often—even when it’s unfair.

Ashcraft and Kirk found that math anxiety functions like a resource-demanding secondary task—preoccupation with math fears consumes working memory capacity, leaving fewer resources for actual computation.
— Ashcraft & Kirk, Journal of Experimental Psychology: General, 2001

Source: Ashcraft & Kirk, Journal of Experimental Psychology: General, 2001

The actual math required

Let’s be concrete. Here’s what a fair split calculation actually requires:

Sample Bill: 5 People

Burger - $18

Salmon - $28

Pasta - $22

Salad - $14

Steak - $45

Shared Apps (3) - $36

Wine (2 bottles) - $90

Dessert (shared) - $16

Subtotal: $269

Tax (8.875%): $23.87

Tip (20%): $53.80

Total: $346.67

To split this fairly, you need to:

Assign individual items

Who had the burger? Who had the salmon? Track 8+ items across 5 people.

Split shared items

The apps were shared by 3 people. The dessert by 4. The wine by 2. Each gets a different denominator.

Calculate individual subtotals

Sum each person’s items + their share of shared items. Five different subtotals.

Distribute tax proportionally

$23.87 × (your subtotal ÷ group subtotal). Requires calculating your percentage of the bill.

Distribute tip proportionally

Same formula, different number. $53.80 × (your subtotal ÷ group subtotal).

Sum final totals

Individual subtotal + tax share + tip share. For each of 5 people. That’s 15 calculations.

This is 30+ arithmetic operations, including percentages, decimals, and division—while remembering which items belong to whom. In a noisy restaurant. With friends watching.

Why we give up

Faced with this complexity, rational people choose the path of least cognitive resistance: “Let’s just split it evenly.”

This isn’t laziness. It’s optimization. Your brain is conserving a scarce resource (working memory) by sacrificing fairness for simplicity.

 80%of people chose to pay individually when given the option
before ordering (Gneezy, Haruvy & Yafe, 2004)—but most won’t
spend the mental effort to calculate it after the fact. 

The tragedy is that equal splits systematically disadvantage certain people: modest orderers, non-drinkers, careful budgeters. They pay the “cognitive load tax”—extra money to avoid the mental work of fair calculation.

The solution: external memory

Cognitive science offers a straightforward solution to working memory limits: offload to external memory.

This is why we write shopping lists instead of memorizing them. Why we use calendars instead of remembering appointments. Why engineers use calculators instead of doing math in their heads.

Writing things down works because it offloads storage from working memory to the page. Each step becomes its own focused task instead of one massive calculation held entirely in your head. Miller’s 7 ± 2 limit stops mattering when external memory handles the rest.

The receipt already contains the information. Every item, every price, every modifier. The cognitive work isn’t gathering data—it’s processing it. A tool that reads the receipt handles the hardest part automatically.

This is what splitty does. Your phone becomes external memory. The camera captures the receipt. OCR extracts the items. You just assign who ordered what. The math is automatic.

From research to design

splitty is designed specifically to work with cognitive limits, not against them:

Working memory: 7±2 items→Show one task at a time. Never require holding multiple calculations.

Cognitive load saturates quickly→No manual entry. Camera captures, OCR extracts, you just assign.

Math anxiety disrupts performance→No visible arithmetic. Totals appear automatically.

External memory outperforms internal→The app is the memory. Persistent, accurate, shareable.

Let your phone do the math.

Your brain wasn't designed for this. Your phone was.