Research

Continuing can be rational.

Imagine:

• —You have already spent €20,000 on a project.
• —Reliable new information indicates that another €2,000 has a high probability of producing €100,000 in value.

The original €20,000 is sunk. But continuing may still make sense because of the future economics.

The fallacy is not: “You invested before and then continued.”

The problem arises when:

The irrecoverable investment itself becomes justification for further investment when it should not affect the forward calculation.

Seven years with a person contain information.

You may know:

• —how conflict is handled
• —whether promises become behaviour
• —whether change has actually happened
• —whether trust exists
• —what recurring patterns look like

That information may legitimately inform predictions about the future.

What does not logically follow is: “Seven years were invested, therefore year eight should happen.”

The useful distinction is not simply past vs future.

Research using sequential decision models has shown that behaviour correlated with previous investment can sometimes arise even in rational agents.

This matters methodologically. Observed persistence plus past investment does not automatically prove irrational sunk-cost reasoning.

We need to know why the history influenced the decision.

This does not mean: “Sunk-cost fallacy is imaginary.”

It means:

Correlation with past investment is not automatically proof of irrationality.

The research does not show that:

• —staying in a long relationship is irrational
• —leaving a long relationship is rational
• —every difficult commitment should be abandoned
• —time invested in something has no emotional meaning
• —past experience should be ignored
• —persistence is automatically evidence of bias
• —every escalation of commitment is caused by sunk costs
• —recognizing sunk cost tells you what decision to make
• —financial decision experiments perfectly generalize to intimate relationships

Past investment can influence future decisions even when that investment cannot be recovered.

But the rational relevance of the past depends on what exactly from the past is influencing the decision.

Ask:

• —“Is this merely an irrecoverable cost?”
• —“Is it information that genuinely changes what happens next?”

The question is not: “How much have I already invested?”