You are on page 1of 13

13. Decoherence and Consistent Histories I.

Quantum Probabilities and Interference

Topics: I. Quantum Probabilities and Interference II. Decoherence III. Consistent Histories

Basic Idea: Classical probability theory is based on classical (Boolean) logic. The probabilities defined by the Born Rule in quantum mechanics are based on quantum (non-Boolean) logic. In particular, they do not satisfy the classical or-addition axiom of classical probability theory.

Classical Probability Theory and the Classical Or-Addition Rule Basic Idea: Begin with a set Ω of simple events (the sample space). Then form a collection F of compound events by taking all possible (classical) logical combinations of the simple events. Then define a probability function Prc that maps elements F to the real interval [0, 1]. Finally slap on a bunch of axioms Prc must satisfy. So: A classical probability theory is given by a triple: (Ω, F, Prc). The probability function p is required to satisfy the following axioms: (C1) (C2) (C3) Prc(∅) = 0 Prc(¬A) = 1 − Prc(A) Prc(A ∪ A') = Prc(A) + Prc(A') − Prc(A ∩ A')
ASIDE: F is a Boolean algebra formed by closing Ω under the operations of set-theoretic complement and countable union.

(C3) is the Classical Or-Addition Rule. It says, "The probability of either A or A' occurring is equal to the sum of the probability of A occuring and the probability of A' occurring, minus the probability of both A and A' occurring." (If A and A' are mutually exclusive, then this latter probability is zero.)

Example: Let Ω = {1, 2, 3, 4, 5, 6} represent the simple events corresponding to all possible results of a single roll of a die. Now form F by taking all possible classical logical combinations of these events: "1", "2", ..., etc., "1 or 2", "1 or 3", ...., etc., "not 1", "not 2", etc... . These correspond to the sets: {1}, {2}, ..., etc., {1} ∪ {2}, {1} ∪ {3}, ..., etc., ¬{1}, ¬{2}, etc... {1, 2} {2, 3, 4, 5, 6}
ASIDE: Again, recall that "or" and "not" form a complete set of classical logic operators (you can form all other relevent operators, like "and", "if, then", etc, just from appropriate combinations of "or" and "not").


L. when V ⊥ W Recall: Linear span ⊕ does not correspond to classical "or".. W are subspaces of H and 0 is the "zero" subspace): (Q1) Prq(0) = 0 (Q2) Prq(V⊥) = 1 − Prc(V) (Q3) Prq(V ⊕ W) = Prq(V) + Prq(W). for instance. 3}) + Prc({3. |ψ〉) = |〈ai|ψ〉|2 . 5}) = Prc({1. Technically. where Prq is defined by Prq(|ai〉. 2. 2. 2. 4. define the probability function to be Prc({i}) = 1/6. 3} ∪ {3. 3} or a value in the range {3. for i = 1. Main Result: Quantum probabilities. 4. So: A quantum probability theory is given by a triple: (H. 5} on a single roll is: Prc({1. 4.Finally. so-defined. it forms what's called a lattice. 2. |ψ〉 are elements of H Recall: This is the probability that the system possesses the value ai of a property represented by an operator with eigenvector |ai〉. Define the probability function Prq on them by the Born Rule. 4. 3} ∩ {3.. when the system is in the state represented by |ψ〉. 2 . ASIDE: Recall from the lecture on quantum logic that the collection of subspaces L is not a Boolean algebra.6. the probability of getting either 1 or 3 on a single roll is given by: Prc({1} ∪ {3}) = Prc({1}) + Prc({3}) − Prc({1} ∩ {3}) = 1/6 + 1/6 − 0 = 1/3 The probability of getting either a value in the range {1. 5}) − Prc({1. Then. do not in general satisfy C3! They do satisfy the following (where V. and is obtained by closing H (thought of as a collection of rays) under the operations of orthocomplement and linear span). Prq). Now form a collection L of subspaces of H (the "compound events") by taking all possible quantum logical combinations of the rays in H (the simple events). 5}) = [Prc({1}) + Prc({2}) + Prc({3})] + [Prc({3}) + Prc({4}) + Prc({5})] − Prc({3} = [1/6 + 1/6 + 1/6] + [1/6 + 1/6 + 1/6] − 1/6 = 5/6 Axiom C3 Axiom C3 Quantum Probability Theory and "Interference" Basic Idea: Replace the classical sample space set Ω with a Hilbert space H. where |ai〉.

which. This is in the subspace V ⊕ W which is the linear span of the subspace V containing the state ψA(x) and the subspace W containing the state ψB(x). Prq(e is at x in state ψB(x)) = |ψB(x)|2 A A B B |ψA(x)|2 + |ψB(x)|2 |ψA(x) + ψB(x)|2 A.Example: 2-slit probabilities and interference |ψA(x)|2 A A B B |ψB(x)|2 A-distribution B-distribution With Slit A open. This is not equal to the sum |ψA(x)|2 + |ψB(x)|2. 3 .or B-distribution (what doesn't happen) Interference distribution (what does happen) With both slits open. according to (C3). represents the probability that the electron either went through slit A or slit B (assuming these are mutually exclusive events). ASIDE: The Born Rule tells us that the state corresponding to the probability distribution |ψA(x) + ψB(x)|2 is the superposition ψA(x) + ψB(x). experiments indicate that the probability that e is located at x is given by |ψA(x) + ψB(x)|2. Prq(electron is located at x in state ψA(x)) = |ψA(x)|2 With Slit B open.

the ray in which |ai〉 is pointing).. |bN〉. The trace has the following properties: Tr(λA) = λTr(A) Tr(A + B) = Tr(A) + Tr(B) Tr(AB) = Tr(BA) where λ is any number The Born Rule can now be rewritten in terms of projection operators: Prq(value of A is ai in state |ψ〉) = |〈ai|ψ〉|2 = 〈ψ|ai〉〈ai|ψ〉 = ∑j〈ψ|P|aj 〉|ai〉〈ai|ψ〉 = ∑j〈ψ|aj〉〈aj|ai〉〈ai|ψ〉 = ∑j〈aj|ai〉〈ai|ψ〉〈ψ|aj〉 = Tr(|ai〉〈ai|ψ〉〈ψ|) = Tr(P|ai 〉P|ψ〉) SO: where ∑jP|a 〉 = 1 j Prq(value of A is ai in state |ψ〉) = Tr(P|ai 〉P|ψ〉) 4 . . First Recall: (1) The projection operator P|ai 〉 = |ai〉〈ai| corresponds to the 1-dim subspace defined by |ai〉 (i. Suppose Q is a linear operator on an N-dim vector space H with orthonormal basis |b1〉. Then the trace Tr(Q) of Q is given by: N Tr(Q ) ≡ ∑ bi Q bi i =1 Note: The trace is just the sum of the diagonal elements of any matrix representation of Q. |b2〉.Let's see how this works in the more general setting of projection operators.. Def.. (2) ∑iP|a 〉 = 1. It turns out that all such representations have this sum in common! So the trace is independent of the basis it's calculated in.e. i Now Consider: The "trace" of an operator.

so the statistical operator is just the projection operator P|ψ〉. Then the statistical ∑kwkP|ψ〉 k (this need not be a projection operator). say |ψ〉. This is the case in which the exact state of the system is known. Aside on terminology: Suppose the exact state of a system is known only to fall within some set {|ψ〉k} to which weights wk can be assigned such that operator W for the system is defined by W = ∑kwk = 1. A pure state is one for which W reduces to a single term. Now: Consider the composite state m-e of a Hardness measuring device and a black electron: |ψ〉 = √{|"hard"〉m|hard〉e + |"soft"〉m|soft〉e} It's statistical operator P|ψ〉 = |ψ〉〈ψ| is given by: P|ψ〉 = {|"h"〉|h〉 + |"s"〉|s〉}{〈"h"|〈h| + 〈"s"|〈s|} = {|"h"〉|h〉〈"h"|〈h| + |"s"〉|s〉〈"s"|〈s| + |"h"〉|h〉〈"s"|〈s| + |"s"〉|s〉〈"h"|〈h|} = {|"h"〉〈"h"|⊗|h〉〈h| + |"s"〉〈"s"|⊗|s〉〈s| + |"h"〉〈"s"|⊗|h〉〈s| + |"s"〉〈"h"|⊗|s〉〈h|} = {P|"hard"〉⊗P|hard〉 + P|"soft"〉⊗P|soft〉 + |"h"〉〈"s"|⊗|h〉〈s| + |"s"〉〈"h"|⊗|s〉〈h|} statistical operator for |"hard"〉m|hard〉e statistical operator for |"soft"〉m|soft〉e interference terms! making sure m-kets belong to m-bras and e-kets belong to e-bras! call it simply √{|"h"〉|h〉 + |"s"〉|s〉} So: Prq(value of A is ai in state |ψ〉) = Tr(P|ai 〉P|ψ〉) = Tr(P|a 〉P|"hard"〉⊗P|hard〉) + Tr(P|a 〉P|"soft"〉⊗P|soft〉) i i + Tr(P|ai 〉|"hard"〉〈"soft"|⊗|hard〉〈soft|) + Tr(P|ai 〉|"soft"〉〈"hard"|⊗|soft〉〈hard|) = Prq(value of A is ai in state |"hard"〉|hard〉) + Prq(value of A is ai in state |"soft"〉|soft〉) + interference terms 5 . the 1-dim subspace defined by |ai〉). P|ψ〉 is the projection operator corresponding to the state |ψ〉 (more precisely. Terminology: The projection operator corresponding to a state is called the statistical operator (alternatively.Again: P|ai 〉 is the projection operator corresponding to the state |ai〉 (more precisely. the 1-dim subspace defined by |ψ〉). A mixed state is one for which W has more than one term. density matrix) for the state.

describe a situation in which a definite measurement outcome occurred. The environment. it does. |soft〉Ε are states of the environment E in which it is correlated with a hard electron and a soft electron. for all practical purposes. for all practical purposes. No collapse: no definite outcome. Decoherence and Interference Claim: Interference effects are destroyed when a system interacts with its environment. environmental interactions effectively "decohere" the state into one in which the observer effectively records a definite measurement outcome. Basic Argument: It is experimentally impossible to distinguish between: (1) The state: √|"hard"〉m|hard〉e|hard〉Ε + √|"soft"〉m|soft〉e|soft〉Ε (2) Either of the states: |"hard"〉m|hard〉e|hard〉Ε or |"soft"〉m|soft〉e|soft〉Ε The reulst of collapse: a definite outcome. There are "interference terms" that are not in general zero. where |hard〉Ε. II. when an observer ends up in an entangled state with a measuring device. it is experimentally impossible to measure such a property. In particular. So (1) and (2) are indistinguishable for all practical purposes! What this is supposed to mean: Whenever the post-measurement state of a composite system is of the form of (1).What this means: The probability associated with the superposed state is not in general the sum of the probabilities associated with its terms. Given that E realistically has a huge number of degrees of freedom. Quantum probabilities do not obey the classical or/addition rule (C3). Recall: To distinguish between (1) and (2). we would need a very complex multi-particle property that (1) possesses and that neither state in (2) possesses. respectively. collapses the entangled superposition. 6 .

The statistical operator P|ψ〉 = |ψ〉〈ψ| for the state in (1) is: P|ψ〉 = {|"h"〉|h〉|Eh〉 + |"s"〉|s〉|Es〉}{〈"h"|〈h|〈Eh| + 〈"s"|〈s|〈Es|} = {|"h"〉|h〉|Eh〉〈"h"|〈h|〈Eh| + |"s"〉|s〉|Es〉〈"s"|〈s|〈Es| + |"h"〉|h〉|Eh〉〈"s"|〈s|〈Es| + |"s"〉|s〉|Es〉〈"h"|〈h|〈Eh|} = {|"h"〉〈"h"|⊗|h〉〈h|⊗|Eh〉〈Eh| + |"s"〉〈"s"|⊗|s〉〈s|⊗|Es〉〈Es| + |"h"〉〈"s"|⊗|h〉〈s|⊗|Eh〉〈Es| + |"s"〉〈"h"|⊗|s〉〈h|⊗|Es〉〈Eh|} = P|"hard"〉⊗P|hard〉⊗P|Eh〉 + P|"soft"〉⊗P|soft〉⊗P|Es〉 + |"h"〉〈"s"|⊗|h〉〈s|⊗|Eh〉〈Es| + |"s"〉〈"h"|⊗|s〉〈h|⊗|Es〉〈Eh|} statistical operator for |"hard"〉m|hard〉e|hard〉E statistical operator for |"soft"〉m|soft〉e|soft〉E interference terms! Now: Take the "partial trace" of P|ψ〉 with respect to the Environment basis |Eh〉. 7 the Environment operators at the end. you just kill states |"hard"〉m|hard〉e or |"soft"〉m|soft〉e. you get zero! When you do the same to the operators P|"hard"〉⊗P|hard〉⊗P|Eh〉 and P|"soft"〉⊗P|soft〉⊗P|Es〉. The "full" basis should be a "3-particle" basis like |"h"〉|h〉|Eh〉. |Es〉. |Es〉: TrE(P|ψ〉) = 〈Eh|P|ψ〉|Eh〉 + 〈Es|P|ψ〉|Es〉 = P|"hard"〉⊗P|hard〉 + P|"soft"〉⊗P|soft〉 Aside: This is called a "partial trace" because we are not using a "full" basis to calculate the diagonal elements of P|ψ〉. This particular "partial trace" results. In this case. |"h"〉|s〉|Es〉. etc (eight such basis vectors). What we're left with is just the statistical operator for the . but a 2-particle operator! Result: "Tracing over the environment" kills the interference terms: They include the operators |Eh〉〈Es| and |Es〉〈Eh|. Basic Claim: "Tracing over the environment" is the mathematical expression of having the environment decohere the entangled superposed state in (1). |"s"〉|s〉|Es〉.Let's see how this works using statistical operators. |"h"〉|h〉|Es〉. not in a number. When you slap |Eh〉 or |Es〉 to both sides of these operators. we're only using a "1-particle" basis |Eh〉.

even when all interference terms are killed by decoherence. we're left with the statistical operator P|"hard"〉⊗P|hard〉 + P|"soft"〉⊗P|soft〉 This is the statistical operator for a "mixed state". the result is either |"hard"〉m|hard〉e or |"hard"〉m|hard〉e. But: The result of a measurement (as given by the Projection Postulate and by our experience) is a definite outcome. P2(t2).. each with equal weight . Of course this doesn't address the basic fact that. . |"soft"〉m|soft〉e}. Pn(tn)} Example: P1(t1) might be P|hard〉 which represents the property "The value of Hardness is hard". III.. However. It's not a weighted sum of them both! Aside: This is the essential criticism of the decoherence approach to the measurement problem. Aside: Projection operators evolve via the Schrödinger dynamics: P(t) = eiHt/P(0)e−iHt/. in such realistic cases. A less severe criticism is the observation that the environment states |Eh〉 and |Es〉 will in realistic situations never be exactly orthogonal to each other. Or it might be P|a〉. This is how the state of a system is represented when its exact form is known only to lie within a set of possible states. we will not kill all the interference terms. A history h is a time-sequence of facts. In this case. the state of the system is either of the pair {|"hard"〉m|hard〉e. represented by time-indexed projection operators: h = {P1(t1). one can show that the "offdiagonal" interference terms that survive in realistic cases have very short life-spans and go to zero very quickly.Does decoherence solve the measurement problem? NO! When we "trace out the environment". which represents the property "The value of the property represented by A is a". the state that results does not describe a definite outcome. In this case. when we "trace over the environment". The way in which the environment correlates itself to a pointer pointing to "hard" may "overlap" the way in which it correlates itself to a pointer pointing to "soft". 9 . Consistent Histories Def 1. It's definitely one of these two alternatives.. Thus.

. 9 ..P2(t2)P1(t1). given it had the value a (of some other operator A) at time t1. Suppose our system is in an initial state |ψ〉.. for any number λ = Tr Pb (t 2 )P a (t1 )P ψ P a (t1 )Pb (t 2 ) since Tr(P|b〉P|a〉P|ψ〉P|a〉) = Tr(P|b〉P|b〉P|a〉P|ψ〉P|a〉) = Tr(P|b〉P|a〉P|ψ〉P|a〉P|b〉) Now extend this to histories of arbitrary length.. Motivation: All the terms inside the trace commute with each other (one of the properties of the trace). |ψ'〉 is the projection of |ψ〉 onto the subspace spanned by |a〉. 2. Aside: A more detailed motivation for Def. call it |ψ'〉. so you can rearrange them into Tr(Pn(tn)Pn(tn).. note that.P2(t2)P1(t1)P|ψ〉P1(t1)P2(t2). To do this. The probability associated with a history h is given by: Prq(h) = Tr(Pn(tn). which can be thought of as the Born Rule for the probability that the system. And since projection operators are idempotent. P|b〉(t2)}.Def 2. and we get Def.P2(t2)P1(t1)P|ψ〉). 2: Consider the simple history h = {P|a〉(t1). this is equal to Tr(Pn(tn). in the state |ψ〉.. It's (normalized) statistical operator is given by Pa P ψ Pa Tr(P a P ψ P a ) P ψ′ = So: Prq(value of B is b in |ψ'〉. Now: Calculate the conditional probability that it has the value b (of some operator B) at t2. then at t2. it will no longer be in the state |ψ〉! Measuring |ψ〉 for a at t1 changes |ψ〉 to a new state.. if it possessed a at t1.. given value of A is a in |ψ〉) = Prq(value of B is b in |ψ'〉)Prq(value of A is a in |ψ〉) = Tr(P|b〉P|ψ' 〉)Tr(P|a〉P|ψ〉) ⎛P P P P ⎞ ⎜ b a ψ a ⎟ ⎟ ⎟ = Tr ⎜ Tr P a P ψ ⎜ ⎟ ⎜ ⎟ Tr( P P P ) ⎟ ⎜ ⎝ a ψ a ⎠ ( ) since Tr(P|a〉P|ψ〉P|a〉) = Tr(P|a〉P|a〉P|ψ〉) = Tr(P|a〉P|ψ〉) = Tr ⎜ ⎜ ⎜ ⎛P P P P ⎞ ⎜ b a ψ a ⎟ ⎟ ⎟ Tr P a P ψ ⎟ ⎟ Tr( P P ) ⎟ ⎜ ⎝ ⎠ a ψ ( ) ) ) ) = Tr Pb P a P ψ P a ( ( ( ) Tr P ( Tr P a P ψ a Pψ since Tr(λA) = λTr(A). has the "historical property" represented by the operator Pn(tn).Pn(tn)) where P|ψ〉 is the statistical operator associated with an initial state |ψ〉.P2(t2)P2(t2)P1(t1)P1(t1)P|ψ〉)...

.. Each set of "exhaustive" facts is represented by a set of projection operators whose sum is the identity. ..... {{P1α1(t1)}.. 3 defines what we'll call below a "fine-grained" history in which all projection operators are 1-dim. Are there histories within a given family that can be assigned classical probabilties? 10 ... . Pn2(tn)} h and h' are distinct histories within the family {{P|ψ〉}. ↑ {P2α2(t2)} ↑ {P1α1(t1)} ↑ . . {P2α2(t2)}.. . P21(t2) P11(t1) h Pn2(tn) . Pi1(ti). PnN(tn) . P2N(t2) . Pn1(tn)} h' = {P|ψ〉. .Def 3. Note: Histories can be embedded in families of histories: t = tn Pn1(tn) . {Pnαn(tn)}}..... ..... . Aside: Def... 2. + PiN(ti) = IN. P11(t1). .. . P12(t1). {P1α1(t1)}.. {Pnαn(tn)}} Note: But: We can assign probabilities to histories within a family by means of Def.. ..... What this means: The projection operators in any such set {Piαi(ti)} represent all the possible values of the property associated with Pi(ti). We still require their sum be the identity. PiN(ti). . Pi2(ti)... .. h' P12(t1) . .. these will be quantum probabilities that exhibit "interference" effects. . A family of histories is a time-sequence of sets of "exhaustive" facts. t = t2 t = t1 t=0 P|ψ〉 {P|ψ〉} h = {P|ψ〉. . {P2α2(t2)}. N = dim of Hilbert space where set {Piαi(ti)} consists of N projection operators... . P22(t2). P22(t2) . There can be "coarse-grained" histories with projection operators with dim > 1.. N such that Pi1(ti) + Pi2(ti) + .. P1N(t1) {Pnαn(tn)} ↑ . P2N(t2).. . .. but now there need not be exactlyl N terms in this sum. where each αi = 1. Since this is based on the Born Rule...

. h or h'....P1(t1)P|ψ〉P1(t1)............ are there histories within a given family that do not "interfere" with each other? These would be histories h.....[PiA(ti) + PiB(ti)].. ... .. . First: Need a general expression for the disjunction....PiB(ti). h'.Pn(tn)) = Tr(Pn(tn). 2 will be classical (i..P1(t1)P|ψ〉P1(t1).. of two histories h.. .Pn'(tn)) = 0 11 . ..PiA(ti). ..PiA(ti). obey the classical OrAddition Rule) just when the interference terms vanish.. .PiB(ti). h' whose probabilities obey the classical Or-Addition Rule: Prq(h or h') = Prq(h) + Prq(h').Pn(tn)) + Tr(Pn(tn). ..Pn(tn)) = Prq(hA) + Prq(hB) + interference terms So: The probabilities assigned to hA and hB by Def.. be given by: hA or hB = {P1(t1). PiA(ti).. PiB(ti)...... obey the classical OrAddition Rule) just when the general interference term vanishes: Tr(Pn(tn).P1(t1)P|ψ〉P1(t1). [Pn(tn) + Pn'(tn)]} The probabilities assigned to h and h' by Def.....e... hA or hB..e. ..PiB(ti). Pn(tn)} Then. Pn(tn)} hB = {P1(t1). for an initial state |ψ〉: Prq(hA or hB) = Tr(Pn(tn). Now: Consider the general case: h = {P1(t1).PiA(ti).. Pn'(tn)} h or h' = {[P1(t1) + P1'(t1)]..P1(t1)P|ψ〉P1(t1). ..Pn(tn)) + Tr(Pn(tn)....... 2 will be classical (i... [Pi(ti) + Pi'(ti)]...... PiA(ti) + PiB(ti). Pn(tn)} h' = {P1'(t1).......[PiA(ti) + PiB(ti)]...PiA(ti).Pn(tn)) + Tr(Pn(tn). Suppose hA and hB are histories that differ only in the property at t = ti: hA = {P1(t1)....P1(t1)P|ψ〉P1(t1)..P1(t1)P|ψ〉P1'(t1).In other words.PiB(ti). Pn(tn)} Now let the history.

Can now say: • Fine-grained histories cannot in general be assigned classical probabilities. Two histories h = {P1(t1). Def. Pn(tn)}. .... 7.. Pn'(tn)} are consistent just when Tr(Pn(tn)...P1(t1)P|ψ〉P1'(t1).P1(t1)P|ψ〉P1'(t1). Pn'(tn)} are decoherent just when Tr(Pn(tn).. ... .P1(t1)P|ψ〉P1'(t1).. . Two histories h = {P1(t1). h' = {P1'(t1).... A decoherent family of histories is a family of histories such that any two histories embeddable in it are decoherent... A consistent family of histories is a family of histories such that any two histories embeddable in it are consistent. 5. such coarse-grained histories "decohere".. What this is supposed to mean: Coarse-graining a family of histories corresponds to "tracing out the environment". 6. Def....Pn'(tn)) → 0. 12 .. Now add the notion of decoherence.... Pn(tn)}. rendering the family approximately consistent... and these get better (more classical) as Tr(Pn(tn).Def. h' = {P1'(t1). • Course-grained histories can be assigned approximate classical probabilities.. (1) h is a fine-grained history just when all projection operators in h are 1-dim. What this means: A consistent family of histories is a collection of histories that defines a classical sample space! You can assign classical probabilities to its members.P1(t1)P|ψ〉P1'(t1). 4. 5..Pn'(tn)) = 0..Pn'(tn)) → 0.. Def..Pn'(tn)) → 0. • As Tr(Pn(tn). The environment interacts with the coarse-grained histories to damp out the interference effects. (2) h' is a coarse-graining of h just when some projection operators in h' are sums of projection operators in h. Def.

Which do we choose in order to calculate the probability of h? (This is the Preferred Basis Problem that Many Worlds faces. All CH does is replace world-talk with history-talk. 13 . • Identifies a way to associate a probability with a history (Def.) 2. 3. 4. • The properties (projection operators) that make up a history evolve only via the Schrödinger dynamics (no Projection Postulate). So it cannot be appealed to in order to reconcile superpositions (or "interfering" histories) with our experience of unique outcomes. 5). General Problem with the Notion of Decoherence "Tracing over the environment" (or "coarse-graining" histories) does not pick out a unique measurement/interaction outcome. but CH is silent as to how to specify the actual history from possible alternatives. • Identifies a condition that picks out those families of histories that are classical (or approximately classical) (Defs. Problems: 1. How are alternative histories within a decoherent family to be interpreted? • Is one such history our actual history and the others just possible? This is our experience. How are alternative decoherent families to be interpreted? Any history h can be embedded in many different mutually incompatible decoherent families (any one of which defines an approximately classical probability space). and adds a criterion for identifying histories that behave "classically".) Problems 1 & 2 Combined: Seem to indicate that CH isn't fundamentally different from Many Worlds. It does not effect a "collapse" of superposed states (or "interfering" histories).Characteristics of the Consistent/Decoherent Histories (CH) Approach • Replaces states of a physical system with histories a physical system. 2). then how are probabilities explained? (This is the Problem of Probabilities that Many Worlds faces. • Are all histories within a decoherent family occurent? If so.