What “given that” means
A probability is always relative to what you know. Before a card is dealt, the chance that it is an ace is 4 out of 52. After you see that the previous card was an ace, the chance that the next one is an ace is different. Conditional probability is the tool for handling that change, and it is written P(A given B): the chance of A once you know B has happened.
The idea is simple even when the notation is not. Learning B removes every outcome in which B is false. What remains is a smaller sample space, and you compute the probability inside it.
The formula and what it means
The standard formula is P(A given B) = P(A and B) ÷ P(B). In words: of all the ways B can happen, what fraction also include A?
For equally likely outcomes this reduces to counting. Count the outcomes where both A and B hold, and divide by the count of outcomes where B holds. You are restricting attention to B’s world and asking how much of it also belongs to A.
Example 1: two dice and a revealed six
Roll two fair dice. Someone looks and tells you that at least one die shows a 6. What is the probability the total is 8?
Start by counting the outcomes that fit what you were told. The pairs with at least one 6 are (6,1), (6,2), (6,3), (6,4), (6,5), (6,6) and then (1,6), (2,6), (3,6), (4,6), (5,6). That is 6 + 5 = 11 outcomes.
Of those, the pairs totalling 8 are (6,2) and (2,6). That is 2 outcomes.
So the probability is 2/11, roughly 18.2 percent. Without any information, the chance of an 8 is 5/36, about 13.9 percent. The revealed six raised it, because 2 of the 11 remaining outcomes reach 8, a larger share than 5 of 36.
The wording of what was revealed matters. “At least one die shows a 6” and “the red die shows a 6” give different answers. In the second case there are only 6 outcomes left, (6,1) to (6,6), and exactly one of them totals 8, giving 1/6. The two statements sound similar and yield different probabilities, so be exact about what you actually learned and how you learned it.
Example 2: cards from a deck
You are dealt one card from a 52-card deck and see that it is an ace. Another card is then dealt. The deck now has 51 cards, of which 3 are aces. The chance the second card is an ace is 3/51, about 5.9 percent, rather than 4/52, about 7.7 percent.
You can also find the probability that both cards are aces from the start: 4/52 × 3/51 = 12/2,652 = 1/221, about 0.45 percent. Dividing by the chance the first card is an ace (4/52 = 1/13) returns 3/51, which confirms the formula: 1/221 ÷ 1/13 = 13/221 = 1/17 = 3/51.
A related card scenario: a dealer turns over a card, and it is not a heart. The chance the next card is a heart is 13/51, since all 13 hearts remain among 51 cards. If the revealed card were a heart, it would be 12/51. Every revealed card changes the remaining composition. This is why tracking what has left the deck is the starting point of deck probability and why counting outs is a conditional calculation in disguise.
Information from players versus information from the deck
Revealed information is not always a physical card. A player’s action can also be evidence. When an opponent makes an unusual bet, the cards they might hold are no longer all equally likely. Turning actions into updated probabilities takes one more step, and Bayes rule describes it. The key point here is that the conditioning principle is the same: restrict attention to what remains possible, then recompute.
Conditioning is not the same as causing
Be careful not to read a conditional probability as a cause. If P(A given B) is higher than P(A), it means B and A tend to occur together, not that B makes A happen. In a game the distinction is usually clear, but in data from many matches it is easy to slip. A player who wins more often when they pick a certain character may simply be a stronger player.
Order does not matter for what you know
A subtle point: it makes no difference to the final result whether the revealed card is the first or second one dealt, as long as you have no information about the other. If two cards are dealt face-down and you later learn the first was an ace, the second card is still one of the remaining 51 equally likely cards, with 3 aces among them. Each unseen card is equally likely to be any card you have not seen. That is why “burn” cards and hidden cards neither help nor hurt your odds when you can’t see them.
Checking your answer
A few habits help avoid mistakes:
- Write down exactly what you were told, in words.
- List or count the outcomes that are consistent with it.
- Count how many of those satisfy the event you care about.
- Divide.
If you can’t list the outcomes, use the formula with counts of combinations. Then sanity-check: the answer should lie between 0 and 1, and if the evidence is closely linked to the event, it should differ noticeably from the unconditional probability.
Conditional probability is also what separates real dependence from the way people feel about streaks. Under a fair random process with independent trials, learning past results does not change the next outcome, so conditioning on them leaves the probability unchanged. Misreading this produces the gamblers fallacy. The difference between dependent and independent situations is the real test: if revealed information changes the odds, the events are dependent, and if it does not, they are independent.