Level 4 — Upper Intermediate · Argument & Evidence Basics
Lesson 31: Critical Reading — Assumptions, Hidden Premises & Logical Gaps
How to find the unstated idea an argument secretly depends on — the hidden premise bridging evidence and conclusion — and how to test whether that bridge can actually carry the weight the author is putting on it.
Learning Objectives
By the end of this lesson, you should be able to:
- Find the hidden premise bridging a piece of evidence to its conclusion, using the test 'what must be true for this conclusion to follow?'
- Tell an assumption (what the argument depends on) apart from an inference (what the reader derives from evidence)
- Separate a necessary assumption, which an argument cannot survive losing, from a merely helpful one
- Recognise ten recurring logical gaps — overgeneralization, scope shift, measurement mismatch, false dilemma, and others — by name and by pattern
- Apply the principle of charity: reconstruct an argument's strongest reasonable form before testing whether it holds up
Introduction
Every argument you've studied so far has an evidence side and a conclusion side. Today's lesson asks about the space between them: what is the author quietly assuming has to be true for the evidence to actually reach the conclusion? That invisible bridge is called an assumption, or hidden premise — and learning to find it is what separates a reader who accepts an argument because it sounds confident from a reader who can tell exactly how far the evidence really carries it.
What Is an Assumption?
The unstated idea an argument depends on
Assumption = something the author hasn't said directly, but has to treat as true for the argument to work.
Evidence: Automation reduced processing time. Conclusion: Automation improved overall operational efficiency.
Does faster processing necessarily mean greater overall efficiency? Not necessarily — automation might have sped things up while also increasing errors, maintenance costs, or training costs. The hidden assumption here: reduced processing time is a valid indicator of improved overall operational efficiency, and no major offsetting costs or problems occurred. Spotting exactly that gap is what this whole lesson trains you to do.
Premise vs. Conclusion
What the argument stands on, and where it's trying to go
Premise — the information the argument is built on. Conclusion — the final point the author wants to establish.
Employees using AI completed tasks 30% faster. Therefore, AI increases employee productivity.
Premise/evidence: employees completed tasks 30% faster. Conclusion: AI increases employee productivity. Between these two sits a bridge that hasn't been stated yet.
Hidden Premise
The invisible bridge, made visible
Continuing the AI example, the hidden premise might be: faster task completion reflects higher productivity. But further assumptions may also be hiding underneath: task quality did not decline; the AI users and non-users performed comparable tasks; the difference was caused by AI rather than by differences between the employees.
EVIDENCE → [hidden assumption] → intermediate conclusion → [hidden assumption] → main conclusion. A critical reader's job is finding every hidden bridge in that chain, not just the first one.
Assumption vs. Inference
What the reader derives, vs. what the argument depends on
The company introduced an automated billing system, and billing errors decreased by 40%.
A reader can reasonably infer: the automated system may have contributed to the reduction in errors — that's an inference, built cautiously from evidence.
But if the author writes Therefore, the automated system improved billing accuracy, they're likely assuming: the reduction in errors was caused mainly by the automated system. That's an assumption — an unstated idea the argument is quietly resting on.
Shortcut: inference = what the reader derives; assumption = what the argument depends on.
"What Must Be True?" — The Core Test
The single most powerful question for finding a hidden assumption
Applying the test to a comparison
Port A processes more vessels per day than Port B. Therefore, Port A is more efficient than Port B.
Does more vessels per day necessarily mean more efficient? Not necessarily — Port A might handle smaller vessels, use more manpower, carry higher operating costs, or create longer waiting times. Hidden assumption: the two ports are being compared under sufficiently similar conditions, and vessels processed per day is an appropriate measure of operational efficiency.
Necessary vs. Helpful Assumptions
Not every assumption carries the same weight
Necessary assumption — the argument cannot survive without it; if it's false, the conclusion is seriously weakened. The new software reduced billing errors. Therefore, it improved billing accuracy. Necessary assumption: the observed reduction in errors is relevant to measuring billing accuracy.
Helpful assumption — strengthens the argument, but the argument doesn't collapse without it. Employees found the new software easy to use is useful supporting context, but not strictly necessary for an efficiency claim.
Necessary = the argument needs it. Helpful = the argument benefits from it.
The Negation Test
Checking whether an assumption is genuinely necessary
Applying the negation test
Possible assumption: faster completion reflects higher productivity. Negate it: faster completion does NOT reflect higher productivity, because quality has substantially declined. If this negation is true, the conclusion (AI increases productivity) falls apart — so the original assumption was doing real, load-bearing work in the argument.
Causal Assumptions
The most common logical gap of all: assuming a cause
After the company introduced AI-assisted scheduling, delivery delays decreased. Therefore, AI-assisted scheduling reduced delivery delays.
After is not because of — new drivers might have been hired, traffic conditions might have improved, delivery volume might have decreased, or route planning might have changed at the same time. Hidden causal assumption: the reduction in delivery delays was caused primarily by the AI-assisted scheduling system rather than by other changes occurring at the same time.
Correlation Sliding Into Causation
X is associated with Y ≠ X causes Y
Employees who use AI tools tend to complete tasks faster. Therefore, AI makes employees more productive.
Alternative explanation: more skilled employees may be more likely to use AI in the first place — so skill could be driving both AI usage and faster performance, with AI itself not the sole cause. A critical reader's honest summary: the evidence shows an association, but it does not by itself establish causation.
Measurement, Scope, Time, and Comparison Assumptions
Four more places a hidden assumption commonly hides
Measurement assumption: The software reduced average processing time by 25%. Therefore, employee productivity increased by 25%. Processing time isn't necessarily productivity — output quality, error rate, complexity, cost, and rework all factor into productivity too. Hidden assumption: processing time is an adequate measure of productivity in this context.
Scope/generalization assumption: A study examined 100 logistics employees in one company. Therefore, AI improves productivity among logistics workers worldwide. Evidence covers 100 employees in one company; the conclusion reaches for logistics workers worldwide. Hidden assumption: the findings from this particular sample and organization can be generalized to workers in other organizations and contexts.
Time assumption: The company was profitable last year. Therefore, its current strategy is successful. Market conditions may have changed. Hidden assumption: the factors responsible for last year's profitability remain relevant to current performance.
Comparison assumption: Port A handles twice as many containers as Port B. Therefore, Port A performs better. Port A might simply be larger, with more cranes and a different cargo mix. Hidden assumption: the relevant conditions and performance measures are sufficiently comparable for the comparison to be meaningful.
Value-Judgment Assumptions
Some hidden premises aren't factual — they're judgments about what's good
The company should automate more operations because automation reduces labor requirements.
The author is assuming: reducing labor requirements is desirable — which isn't automatically true. An organization might have other goals: employment, service quality, resilience, employee development, social responsibility. A critical reader watches for value assumptions just as carefully as factual ones.
Ambiguous Words as a Source of Logical Gaps
A vague key term can hide the whole gap
Words like effective, efficient, successful, better, improved, significant, useful, optimal, reliable, high-quality deserve extra scrutiny. The new system is more effective. — effective in what sense? Cost? Speed? Accuracy? User satisfaction? Reliability? Security? An argument's conclusion can look solid while actually resting on an undefined key term.
Ten Common Logical Gaps
| Pattern | Shape of the gap |
|---|---|
| Missing causal link | X happened → therefore X caused Y |
| Overgeneralization | A few cases → conclusion about everyone |
| False equivalence | Two things share one feature → treated as equivalent |
| Unsupported comparison | A is better than B, but the comparison criterion is unclear |
| Correlation → causation | X associated with Y → X causes Y |
| Anecdote → generalization | One person's experience → treated as generally true |
| Scope shift | One company's evidence → conclusion about an entire industry |
| Time shift | Past evidence → conclusion about the current situation |
| Measurement mismatch | Metric A improved → therefore broader concept B improved |
| False dilemma | Presented as "either A or B" when C, D, and more are also possible |
"Therefore" Doesn't Guarantee Valid Logic
Therefore reports the author's claimed connection — not a verified one
Processing time decreased significantly. Therefore, overall operational performance improved.
Seeing therefore doesn't mean the logic is automatically correct — it only shows you what connection the author is claiming. Your job is always: is this connection actually justified?
True Evidence ≠ Supported Conclusion
A fact can be real while the conclusion built on it isn't
Fact: The new system reduced processing time by 20%. This can be completely true. Conclusion: The new system improved the entire organization. This does not necessarily follow. Evidence can be true while the conclusion resting on it is insufficiently supported — always keep these two judgments separate.
The Principle of Charity
Reconstruct the strongest version before you test it
Critical reading doesn't mean hunting for the worst possible reading of an argument. First: reconstruct the strongest reasonable version of the argument. Then: test it.
Bad critical reading: "This argument is wrong because the author left out a small detail." Good critical reading: "The argument is plausible, but its conclusion depends on an unstated assumption about measurement and alternative explanations." The second version is what mature, fair-minded critical reading actually looks like.
A Full Argument Map
Applying every tool from this lesson at once
The port introduced an AI-based vehicle-tracking system. After implementation, average truck waiting time fell by 30%. Management therefore concluded that the system significantly improved overall port efficiency.
Evidence: average truck waiting time fell by 30%. Main claim: the system significantly improved overall port efficiency. Hidden assumption 1: reduced truck waiting time is a meaningful indicator of overall port efficiency. Hidden assumption 2: no major negative effects offset the improvement. Hidden assumption 3: the reduction was caused primarily by the new system. Alternative explanations: traffic volume may have decreased; additional gates may have opened; staffing may have increased. Logical gap: evidence measures one operational metric, but the conclusion claims improvement in overall efficiency — the conclusion may be too strong for the evidence.
Better conclusion: the system appears to have improved one important aspect of port operations, although its effect on overall efficiency requires additional evidence.
A Ten-Step Critical Reading Workflow
Working through any difficult argumentative paragraph
Deep Reading Example
A logistics passage, worked through the full workflow
A logistics company introduced an AI-based scheduling system across three distribution centers. Within six months, average delivery delays fell by 18%. Managers therefore concluded that the system had substantially improved operational efficiency. Although employee training was conducted during the same period, management attributed most of the improvement to the AI system because the largest reduction occurred at the center that used the system most extensively.
Sentence 1 — background/setup. Sentence 2 — evidence/ observation: an 18% reduction. Sentence 3 — conclusion: "18% reduction → substantially improved operational efficiency" — where's the bridge? Sentence 4 — the author names its own alternative explanation (training) and offers a dose-response pattern (more AI usage → larger improvement) as supporting reasoning — useful, though still not conclusive.
Hidden assumptions here: delivery delays are an adequate indicator of operational efficiency; the reduction was not primarily caused by employee training; other operational changes didn't explain it; the centers were sufficiently comparable; greater AI usage is meaningfully associated with the larger improvement.
What a strong critical reader actually says
Not "the managers are wrong." Instead: "The evidence is consistent with the claim that the AI system improved operational performance, but it does not by itself establish that the system was the primary cause of the improvement. The concurrent employee training and other possible operational changes remain alternative explanations. In addition, the reduction in delivery delays may not fully capture overall operational efficiency." This is what advanced academic reading sounds like — not a verdict, but a calibrated judgment.
Vocabulary in Context
take for grantedphraseneutral
to accept something as true without examining or questioning it (পরীক্ষা না করে সত্য ধরে নেওয়া)
Often with take it for granted, taken for granted, easily taken for granted
“The plan takes for granted that every supplier will deliver on time, which the record does not justify.”
underlying assumptionnoun phraseformal
an unstated idea beneath the surface of an argument, on which the argument's logic depends (অন্তর্নিহিত অনুমান)
Often with challenge the underlying assumption, hidden underlying assumption, flawed underlying assumption
“The underlying assumption of the forecast is that fuel prices will not rise, and the whole budget depends on it.”
implicitadjectiveformal
suggested or understood without being directly stated (অন্তর্নিহিত/অব্যক্ত)
Often with implicit assumption, implicit criticism, implicit in
“The memo contains an implicit criticism of the previous management team without naming anyone.”
explicitadjectiveformal
stated directly and clearly, leaving nothing to be inferred (স্পষ্টভাবে বলা)
Often with explicit instructions, explicit warning, explicitly state
“The tender document is explicit about penalties for late delivery.”
logical gapnoun phraseformal
a missing link in reasoning where a conclusion doesn't fully follow from its evidence (যুক্তির ফাঁক)
Often with a logical gap, fill a logical gap, gap in the argument
“The report has a logical gap: it shows that costs rose but never explains why the terminal is to blame.”
follow fromphraseformal
to be a logical consequence of something stated earlier (যুক্তিসঙ্গতভাবে উদ্ভূত হওয়া)
Often with follow from the evidence, not necessarily follow, conclusion follows from
“It does not follow from one late delivery that the supplier is unreliable.”
rule outphrasal verbneutral
to eliminate a possibility as unlikely or impossible (বাদ দেওয়া/অগ্রাহ্য করা)
Often with rule out the possibility, cannot rule out, rule out a cause
“Investigators have not ruled out human error as a contributing cause of the collision.”
generalizeverbformal
to apply a finding from a specific sample or study to a broader population or context (সাধারণীকরণ করা)
Often with generalize from, cannot generalize, safely generalize
“It would be unsafe to generalize from a single pilot site to the entire network.”
adequateadjectiveformal
sufficient or satisfactory for a particular purpose (যথেষ্ট/উপযুক্ত)
Often with adequate evidence, wholly adequate, adequate safeguards
“The regulator found that the operator had not put adequate safeguards in place.”
overstateverbformal
to describe something as more significant, certain, or extensive than the evidence actually supports (অতিরঞ্জিত করা)
Often with overstate the case, overstate the risk, significantly overstate
“The press release overstates the benefits of the new system by ignoring its installation costs.”
Guided Reading Practice
Read this passage once for its overall claim, then find the hidden assumption bridging its evidence to its conclusion before checking the notes underneath it.
A shipping line introduced a new digital documentation platform across its regional offices. Within the first quarter, the average time to process a bill of lading fell from three days to one day. Regional management concluded that the platform had transformed the company's competitiveness. However, the same quarter also saw the company simplify several customs-clearance procedures and hire additional documentation staff at its busiest offices.
Evidence: processing time fell from three days to one day. Conclusion: the platform transformed the company's competitiveness — a much broader claim than the evidence directly supports. Hidden assumptions: faster documentation processing is a meaningful driver of overall competitiveness; the reduction was caused primarily by the platform rather than by the simplified procedures or additional staff; a faster process necessarily translates into a competitive advantage in the market. Logical gap: scope shift — one process metric (documentation time) is being used to support a claim about the company's entire competitive position.
Golden Rule
Golden Rule
Before accepting a conclusion, ask what must be true for the evidence to actually support it — and check whether that unstated assumption can survive being questioned.
Lesson Summary
Today's lesson trained you to see the part of an argument that never gets written down: the assumption bridging evidence to conclusion. You practised the core test — what must be true for this conclusion to follow? — alongside the negation test for checking whether an assumption is genuinely necessary. You learned to recognise causal, measurement, scope, time, comparison, and value-judgment assumptions as distinct categories, and to name ten recurring logical gaps by pattern rather than by vague unease. Throughout, the principle of charity kept you honest: reconstruct the argument's strongest form first, then test it — never attack a weaker version than the one the author actually made.
The habit to carry forward
Every time a conclusion follows a "therefore," pause and ask: what unstated idea does this argument need me to accept for that "therefore" to actually work? That question is the whole of today's lesson, compressed into a habit.
Practice: Test What You've Learned
Work through every question yourself before checking anything.
Before you start
Write out each hidden assumption as a complete sentence before judging it — a vague sense that "something's missing" is much easier to correct than a clearly stated, testable assumption.
Part A — Claim and Evidence
- Employees using AI completed tasks 30% faster. Therefore, AI increases employee productivity. Identify the evidence and the conclusion.
Show answer
The evidence is that employees using AI completed tasks 30 percent faster. The conclusion is that AI increases employee productivity. Between them sits an unstated bridge connecting speed to productivity. - Write one hidden assumption this argument depends on.
Show answer
The argument assumes that faster task completion reflects genuinely higher productivity, meaning the quality of the work did not decline and the speed gain was not offset by other costs, such as increased errors or extra training time.
Part B — Hidden Assumption
- Port A processes more vessels per day than Port B. Therefore, Port A is more efficient than Port B. Using the "what must be true?" test, write the hidden assumption in full.
Show answer
For this conclusion to follow, we must assume that the two ports are being compared under sufficiently similar conditions, such as vessel size, staffing levels, and equipment, and that vessels processed per day is itself an appropriate measure of operational efficiency rather than of size or resources alone. - Apply the negation test to that assumption: what happens to the conclusion if the assumption is false?
Show answer
If the assumption is negated, so the ports are not comparable, for instance if Port A simply handles smaller vessels with far more staff and cranes, then a higher vessel count no longer demonstrates efficiency. The conclusion that Port A is more efficient collapses, which shows the original assumption was doing real, load-bearing work.
Part C — Necessary vs. Helpful
- The new software reduced billing errors. Therefore, it improved billing accuracy. Identify the necessary assumption.
Show answer
The necessary assumption is that the observed reduction in errors is relevant to, and a valid indicator of, billing accuracy. Without this connection, a drop in errors and an improvement in accuracy have no guaranteed relationship, and the conclusion has nothing to stand on. - Employees found the new software easy to use. Explain why this is a helpful assumption rather than a necessary one for the claim in question 5.
Show answer
Ease of use might help explain why errors fell, but the accuracy conclusion could still stand even if employees found the software difficult, as long as errors actually decreased. Since the argument does not collapse without it, it only strengthens the case rather than being required by it.
Part D — Logical Gap and Assumption vs. Inference
Read this passage for questions 7-10: The port introduced an AI-based vehicle-tracking system. After implementation, average truck waiting time fell by 30%. Management therefore concluded that the system significantly improved overall port efficiency.
- Which of the ten logical-gap patterns from this lesson best describes the jump from "waiting time fell" to "overall efficiency improved"?
Show answer
This is a measurement mismatch: one specific, narrow metric (truck waiting time) improved, but the conclusion claims improvement in a much broader concept (overall port efficiency) that waiting time alone does not establish. - Name two alternative explanations for the 30% reduction.
Show answer
Traffic volume at the port may have decreased independently of the tracking system, or additional gates may have opened and extra staff may have been added during the same period, speeding up processing regardless of the new system. - Write a reasonable inference a careful reader could draw from this passage.
Show answer
A careful reader could reasonably infer that the vehicle-tracking system may have contributed to the reduction in truck waiting time, without claiming it was the sole or primary cause. - Write an assumption the argument itself depends on. How does it differ from your answer to question 9?
Show answer
The argument assumes that the reduction in waiting time was caused primarily by the tracking system and that it reflects genuine improvement in overall efficiency. This differs from the cautious inference in question 9 because an inference is what a reader carefully derives from the evidence, while an assumption is the stronger, unstated claim the argument needs accepted for its confident "therefore" to actually hold.
Part E — Full Paragraph Critical Analysis
Return to this lesson's deep reading passage about the logistics company that introduced AI-based scheduling across three distribution centers.
- List all five hidden assumptions identified in the lesson.
Show answer
1) Delivery delays are an adequate indicator of operational efficiency. 2) The reduction was not primarily caused by the concurrent employee training. 3) No other operational changes explain the improvement. 4) The three distribution centers were sufficiently comparable. 5) Greater AI usage is meaningfully associated with the larger improvement at the center that used it most. - Which of these five would you classify as necessary, and which as merely helpful? Justify your choice for at least two of them.
Show answer
"Delivery delays are an adequate indicator of operational efficiency" is necessary, because if this is false, the entire efficiency conclusion has no measurement to stand on. "The reduction was not primarily caused by employee training" is also necessary, because if training were the real cause, crediting the AI system would be unfounded. "The centers were sufficiently comparable" and "greater AI usage is associated with the larger improvement" are more helpful than necessary, since they strengthen confidence in the causal story but the core claim could survive some unevenness between centers, as the dose-response pattern is only supporting evidence rather than the argument's foundation. - Write your own critical-reader response to this passage, following the model given in this lesson (not "the managers are wrong," but a calibrated judgment).
Show answer
The evidence is consistent with the claim that the scheduling system improved delivery performance, but it does not by itself establish that the system was the primary cause. Concurrent employee training remains a plausible alternative explanation, and the dose-response pattern, while suggestive, does not rule out other operational differences between centers. An 18 percent reduction in delivery delays is real, but it may not fully capture the broader concept of operational efficiency the managers are claiming. - Apply the principle of charity: state the strongest reasonable version of the managers' argument before you critique it.
Show answer
The strongest reasonable version is that a company-wide AI scheduling system was followed by a substantial, sustained 18 percent reduction in delivery delays across three centers, with the largest gains concentrated at the center that used the system most heavily, a dose-response pattern that at least suggests the system made a genuine causal contribution even though training occurred at the same time.
Lesson 32 — Counterfactual Thinking, Causal Reasoning & "What Would Happen If…?": pushing causal reasoning one step further by asking not just what happened, but what would have happened if the supposed cause had never occurred.