Found Phys (2008) 38: 1073–1081

DOI 10.1007/s10701-008-9252-y

Interpreting Quantum Interference Using a Berry’s

Phase-like Quantity

M.J. Rave

Received: 27 June 2008 / Accepted: 15 October 2008 / Published online: 22 October 2008
© Springer Science+Business Media, LLC 2008

Abstract We show that quantum interference can be interpreted in terms of a phase

invariant quantity, not unlike the Berry’s phase. Under this interpretation, closed
loops in time become fundamental quantum entities, and all quantum states become
periodic. Decoherence is then seen to occur naturally as a consequence. This formal-
ism, although counterintuitive, provides another useful way of assigning meaning to
quantum probabilities and quasi-probabilities.

Keywords Berry’s phase · Quantum interference · Quantum probability ·

Decoherence · Closed timelike curves

1 Introduction

To many, quantum interference (QI) is the most mysterious aspect of quantum me-
chanics; in Feynman’s words, it “contains the only mystery” [1]. The implication is
that all of the non-intuitive aspects of quantum mechanics (such as wave-particle du-
ality, non-commuting observables, the EPR “paradox” [2], or Bell’s inequality [3])
are manifestations of a single non-classical phenomenon, namely QI. It has become
common to follow Feynman’s lead and say that there is no way to explain QI in
a “classical way” [1]; by this it is meant that QI involves mixed probabilities (cross-
terms) that do not appear in classical probability theory. However, recent work [4] has
shown that certain decidedly classical systems can exhibit superficially non-classical
probabilities and pseudo-QI effects. Thus it becomes natural to reexamine QI in an
attempt to reinterpret non-classical terms. Clearly, an understanding of why such non-
classical probabilities show up in quantum systems is important in developing any
meaningful interpretation of quantum mechanics itself. In the words of Mermin, “an

M.J. Rave ()

Department of Chemistry and Physics, Western Carolina University, Cullowhee, NC 28723, USA
e-mail: mrave@[Link]
1074 Found Phys (2008) 38: 1073–1081

acceptable notion of probabilities as objective properties of individual systems” may

“sweep all the puzzles of quantum mechanics” under the rug [5].
Part of the perceived difficulty in interpreting quantum mechanics lies in the struc-
ture of quantum mechanics itself: what concepts, if any, should be considered fun-
damental? For example, are state vectors more fundamental than operators? At first
the question seems moot, since the Schrödinger and Heisenberg pictures are math-
ematically equivalent (up to a change of basis). But consider that all observables
correspond to Hermitian operators: does this mean that such operators have a special
place in the hierarchy of quantum ideas?
The concept of the Berry’s phase (BP) [6] has shed new light onto such philo-
sophical questions. In 1984 Berry showed that although the phase of a given state
vector (ket) cannot have physical meaning, a phase invariant quantity (now called the
Berry’s phase) can always be constructed in the Hilbert space of the state vectors,
and this phase is measurable. Berry’s idea came as a revelation in the mid-1980’s, for
the standard dogma of the time was that for a quantity to be measurable, it must be
the eigenvalue of some Hermitian operator. Berry showed that Berry’s phase is mea-
surable, but occurs as a result of the topology of the Hilbert space in which the BP
resides. Hence there are no operators that can be associated with the BP. In Resta’s
words, Berry’s phase is an “exotic” observable [7]. Being observable, then, is there
any reason to place the BP on par with state vectors and/or Hermitian operators as
quantum mechanical constructs?
The outline of the rest of this paper is as follows. First, we explain the BP, and
why the notion of closed loops is so integral to the BP concept. Second, we discuss
quantum probability and give an operational definition of QI. Third, we put the two
ideas together, showing how a closed loop paradigm allows one to interpret QI in
a natural—if not precisely classical—way. We conclude with discussions of deco-
herence, closed loops in time, and the connection to the path integral approach to
quantum mechanics.

2 Berry’s Phase

Berry’s phase is usually introduced within the context of a quantum system governed
by a parametric Hamiltonian H(x) [7]. (This idea is convenient for didactic purposes,
but is not required for the BP to manifest itself.) For now we will define the BP as a
phase shift in a system’s state vector |(x) that appears after the system has traced
a closed loop in the space of some parameter x. Consider a system that evolves cycli-
cally as |(x1 ) → |(x2 ) → |(x3 ) → |(x1 ). (We will refer to this ordered
sequence of states as a “path” even though it involves a discrete number of states.
The path is considered closed because the final state of the system is the same as the
initial state.) For this path the BP (γ ) is defined to be [7]
γ = −Im ln[(x1 )|(x3 )
(x3 )|(x2 )(x2 )|(x1 )]. (1)
The most important feature of this definition is that γ is phase invariant (sometimes
called “gauge invariant”). By this we mean that individual phase choices for partic-
ular kets don’t matter; the BP still comes out the same for a given path. This can
Found Phys (2008) 38: 1073–1081 1075

be seen by making the “gauge transformations” |(x1 ) → eiθ1 |(x1 ), |(x2 ) →
eiθ2 |(x2 ), |(x3 ) → eiθ3 |(x3 ) and recalculating γ .
Because it is invariant to phase choice, the discrete BP (and its continuum limit
counterpart) is a measurable quantity. For example, in systems of Bloch electrons the
presence of the BP is seen as either an electron self-interference effect ([8, 9]) or as
a half-integer quantum Hall effect [10]. The BP appears in these particular systems
because of the torus topology of the Brillouin zone; k’s (Bloch pseudomomenta)
that differ by a reciprocal lattice vector are considered equal. Similarly, in all other
situations in which a BP appears (electric polarization [11], spin-wave dynamics [12],
or even molecular machines [13]) the state vectors traverse a closed loop in the space
of some parameter. The importance of such closed loops motivates further research
into placing the BP into the structure of quantum theory.

3 Quantum Probabilities

Recent work [14] has shown that there is a link between the BP and quantum prob-
ability. Specifically, in employing a consistent histories [15] formulation of quantum
mechanics, one finds that the BP becomes “the main building block of the decoher-
ence functional” [14], which itself contains complete information about a quantum
system. Put another way, for each quantum mechanical history there can be associ-
ated (mapped) a unique BP. The authors of [14] show that “all physically relevant
amplitudes” are based on the BP, and conclude that the presence of complex phases
leads to non-classical probability terms (i.e. QI). We do not disagree with this result
but suggest that it is not complex phases per se that lead to QI, but the presence of
closed loops in time. We also suggest that the significance of such closed loops may
lead to a paradigm shift in the analysis of quantum transitions.
For the purposes of our discussion, we will define QI through an examination of
probability. We say that a system lacks QI if the system always obeys the addition
law [16]
P (A1 ∪ A2 ) = P (A1 ) + P (A2 ), (2)
with A1 and A2 being mutually exclusive events, and P (A1 ∪ A2 ) being the union of
these events. In contrast, a system with QI is immediately identifiable by the presence
of equalities such as

P (A1 ∪ A2 ) = P (A1 ) + P (A2 ) + f (P (A1 ), P (A2 )), (3)

where f (P (A1 ), P (A2 )) is, in general, non-zero, although it may be negative.

In analogy with the mathematics of interfering waves, we call mixed terms like
f (P (A1 ), P (A2 )) “interference terms” [1].
Note that this definition of quantum interference is an operational one; we do not
enter into any debate between frequentists and Bayesians [17] as to the epistemo-
logical meaning of probability. Nor do we refer to past and/or future events in order
to give meaning to probabilities, as some have done (e.g. [18]). Our goal is simply
to show that quantum interference terms become naturally explicable (and in one
view cease to be “interference” terms at all) if one looks at “closed loops in time” as
fundamental quantum entities.
1076 Found Phys (2008) 38: 1073–1081

4 Product of Amplitudes

We represent state vectors by kets in a complex vector space. In such a space | and
eiθ | are said to correspond to the same physical state, since they lead to identical
expectation values for identical Hermitian operators. Now consider two distinct states
|1  and |2 . The inner product φ21 ≡ 2 |1  is called a probability amplitude,
since the probability of the transition 1 → 2 is taken to be |2 |1 |2 . Since
φ21 ∈ C, we can write the amplitude in modulus/phase form as

φ21 = r21 e−iθ21 . (4)

The phase difference θ21 = − Im ln2 |1  is not a “Berry’s phase” since we do not
yet have a closed loop.
Now assume that we have three states that evolve cyclically, and calculate the
(phase invariant) BP as in (1). In terms of amplitudes, we have

γ = − Im ln φ13 φ32 φ21 . (5)

Because the BP is independent of phase choice, so too is the product of amplitudes

φ13 φ32 φ21 . This suggests the definition


N
Ŵ ≡ φ1N · · · φ32 φ21 = φi+1,i , (6)
n=1

where it is understood that the loop eventually closes upon itself so that N + 1 → 1.
So defined, Ŵ is not a phase, but (like the BP) it is phase invariant. Moreover, Ŵ
(what we call a “product of amplitudes”) has a natural interpretation in terms of
probabilities.
To see how this is so, we make the following observation: the probability of going
from an initial state |i to a final state |f  is the sum of all the Ŵ’s that include |i
and |f . Explicitly,

Pf i = Ŵ(S), (7)
S
where the S’s represent all possible closed paths that include |i and |f . Note that
since Pf i must be real and non-negative, the Ŵ’s themselves must also then be real.
Indeed, it straightforward enough to see that although the individual factors (ampli-
tudes) that make up a particular Ŵ are complex, the Ŵ’s themselves are real.
Since we have a probability Pf i that is the sum of real terms, it is tempting to
interpret each individual Ŵ as a probability (perhaps as the probability of a particular
closed path). Note, however, that there is the possibility of a particular Ŵ being neg-
ative. The idea of negative probabilities in quantum mechanics was first proposed by
Dirac in 1942 [19], but cannot in any way be considered traditional. For this reason,
we refer to Ŵ’s as being “quasi-probabilities”. Our interpretation of (7) is then: the
probability of going from an initial state |i to a final state |f  is the sum of all the
quasi-probabilities that include |i and |f , each of which corresponds to a unique
closed loop.
Found Phys (2008) 38: 1073–1081 1077

To see how this formalism works, suppose a state |i goes to a state |f  via only
one possible (intermediate) route |1:

|i → |1 → |f . (8)

Traditionally we would write, for the amplitude of an |i → |f  transition via |1,

f |i = f |11|i = φf 1 φ1i ≡ φf i,1 , (9)

so that the probability of such a transition is simply Pf i = |φf i,1 |2 . By our “Ŵ-rule”,
we get the same result: there is only one possible closed loop,

|i → |1 → |f  → |1 → |i, (10)

and we get for the total probability

Pf i = i|11|f f |11|i = φf∗ i,1 φf i,1 = |φf i,1 |2 , (11)

as before. (Note that there are not yet any interference terms as in (3).) We resist the
temptation at this point to interpret the meaning of such a closed loop; for now we
consider it a convenient mathematical device.
To get “interference”, we require there to be two means to go from |i to |f : via
|1 or |2. Traditionally, one finds that

f |i = f |11|i + f |22|i, (12)

giving a probability of

Pf i = |f |i|2 = |φf i,1 |2 + |φf i,2 |2 + φf i,1 φf∗ i,2 + φf∗ i,1 φf i,2 (13)

for the |i to |f  transition.

Observe that interference terms are now present: these are the mixed terms
φf i,1 φf∗ i,2 and φf∗ i,1 φf i,2 , which supposedly have no easy interpretation. Indeed, if
we insist on reading Pf i as a “probability of a state i evolving into state f ” then the
interference terms remain entirely inexplicable. After all, routes 1 and 2 are mutually
exclusive, so there is no obvious reason why Pf i should depend in any way on such
mixed terms. Since (2) does not hold, we are left with interference as the epitome of
“quantum weirdness” [20].
Our “Ŵ-rule”, however, provides an equivalent way to calculate Pf i . We have four
possible closed loops,

|i → |1 → |f  → |1 → |i,

|i → |2 → |f  → |2 → |i,
(14)
|i → |1 → |f  → |2 → |i,
|i → |2 → |f  → |1 → |i,

with the respective quasi-probabilities |φf i,1 |2 , |φf i,1 |2 , φf i,1 φf∗ i,2 , and φf∗ i,1 φf i,2 .
Thus the total probability Pf i is found to be the same as in (13). Perhaps this is
1078 Found Phys (2008) 38: 1073–1081

mathematical coincidence, but the presence of the phase invariant BP-like quantity Ŵ
suggests another explanation. It is natural to conclude from this analysis that closed
loops are in some sense more fundamental than transitions between states. The au-
thors of [14] hint at this, but shy away from our hypothesis. Going one step further,
we suggest the following: that in deciding on the probability of a transition |i to |f ,
one can simply choose from all possible closed loops that contain |i and |f ; there
is a unique quasi-probability associated with each such loop. In other words, Pf i
can be considered to be the sum of four mutually exclusive alternatives, as opposed
to being a sum of two alternatives with two ad hoc interference terms tacked on. In
the language of the many-worlds viewpoint of quantum mechanics [23, 24], we asso-
ciate a different “universe” with each closed loop in time; for each “universe” there
is a well-defined quasi-probability.

5 Quantum Decoherence

Interestingly, a closed loop viewpoint can also provide insight into the phenomenon
of quantum decoherence [25]. By quantum decoherence we mean the process by
which mixed probability terms (such as φf i,1 φf∗ i,2 , and φf∗ i,1 φf i,2 ) vanish when a
system interacts with its environment. This is not full-fledged wave function collapse
à la the Copenhagen interpretation; decoherence (and its consequence, einselection)
produces an effective wave function collapse [25]. Essentially, the environmental in-
teraction (a “measurement”) rules out certain mixed amplitudes, and we (macroscop-
ically) see the disappearance of QI.
Consider again the quantum system that evolves from |i to |f  via one of the
intermediate states |1 or |2. There are two distinct possibilities for state |f : either
1. |f  does not depend in any way on the intermediate state by which the system
“arrived” at |f , or
2. |f  does depend upon the intermediate state. In such a case we write |f (1) or
|f (2) to indicate the “route” taken.
The phase invariance of the product of amplitudes suggests that we look at these
systems in terms of closed loops and not just as quantum transitions. In the first
case, there are four possibilities, denoted by 11∗ , 12∗ , 21∗ , and 22∗ , with an asterisk
indicating a time-reversed route. Let us consider such closed loops to be fundamental
quantum entities. (An adherent of the many-worlds viewpoint would say that each of
these four possibilities represents four possible universes that are consistent with the
“fact” of |f .) Notice that there is no contradiction in supposing (for example) that a
system could evolve as |i → |1 → |f  → |2 → |i, since the state |f  contains no
information about the path taken.
In the second case (with |f (1) or |f (2)), however, the mixed routes 12∗ and 21∗
lead to time reversal “paradoxes”. A state such as |f (1) or |f (2) which contains
(as one of its properties) information about how the state was achieved, cannot then
evolve back into state |i by a different path. That is, a loop such as |i → |1 →
|f (1) → |2 → |i is clearly impossible. The only choices then available are the
closed loops 11∗ and 22∗ ; as a consequence the QI probability terms have vanished.
Found Phys (2008) 38: 1073–1081 1079

Fig. 1 Two ways to visualize a

closed loop in time. The first
way, with a forward time t and a
“backward time” t ∗ , seems
nonsensical. The second way is
mathematically equivalent

In effect, the making of any measurement rules out any mixed closed loops, which
do not then enter into any probability calculations.
This is not to say that a measurement (and consequent decoherence) actively
changes a state that does not depend on intermediate states to a different state that
does. Rather (from a many-worlds viewpoint) a measurement rules out the possibil-
ity that we are in a mixed-loop universe. For the above system, if all we know is that a
transition |i → |f  takes place, then there is no way in general to choose among the
four possibilities 11∗ , 12∗ , 21∗ , and 22∗ ; each is quite possibly the universe we are
in. But if a state depends in some way upon the path taken, then (upon finding such
a state) we must be in either the “11∗ ” universe or the “22∗ ” universe. Decoherence
has been achieved by a simple selection from the available “legal” closed loops. The
state |f  must be executing a self-retracing round trip since an “exotic” round trip
(that never retraces itself) would contradict what we know about |f .

6 Possible Interpretations

We now make a few remarks regarding this discussion. Firstly, all quantum transitions
can be couched in the language of our product of amplitudes (Ŵ), and this quantity
(like the Berry’s phase) is always phase invariant. Therefore, regardless of whether
or not we believe that the systems in question are “actually” traversing closed loops
in time, we have yet another mathematical framework in which to interpret quantum
probabilities. But it is precisely the invariant nature of Ŵ that suggests we go further:
using closed loops in time allows QI to be interpreted in terms of probabilities, so
perhaps such loops are important. In the language of the Berry’s phase, perhaps time
itself should be considered a parameter of a state’s governing Hamiltonian, and prob-
abilities should be calculated based upon the assumption that all states eventually
“loop back on themselves”.
Of course, treating time this way is problematic, for it is not obvious what a closed
loop in time means. (Indeed, such a supercyclic time requires us to abandon principles
of causality: future events can be the “cause” of something in the past [26].) But for
visualization purposes, there is no mathematical difference between a Feynman-like
diagram in which states loop back onto themselves and a linear diagram that shows
states being periodic. Both evolutions yield identical BP’s and identical Ŵ’s.
In light of this, we speculate that the appearance of closed timelike curves [27, 28]
in certain solutions to Einstein’s field equation (such as in the Kerr metric [29]) are
not mere artifacts. Rather, they might provide evidence that the closed loop viewpoint
of time is not merely a mathematical convenience, but may have some independent
significance. This ties in with recent speculation that the universe itself may be time
symmetric on large (cosmological) scales [30].
1080 Found Phys (2008) 38: 1073–1081

Secondly, one may ask how this interpretation differs from Feynman’s path in-
tegral formulation [22] of quantum mechanics. The answer is that it differs only in
viewpoint. According to the path integral prescription, the probability amplitude of a
particular transition is calculated by adding up all of the individual amplitudes (his-
tories) that include the initial and final states, just as in (12). The probability of this
transition is then postulated to be the squared modulus of this total amplitude.
In light of the phase invariant product of amplitudes, however, we see that there
is another viewpoint. We consider as fundamental not only histories, but futures, and
include all such history/future combinations (loops). For each such loop, there is a
unique quasi-probability; the total probability of the transition is the sum of these
quasi-probabilities. When doing this, mixed terms (ostensibly due to interference)
are seen to be mutually exclusive alternatives associated with exotic round trips. If
taken at face value, this suggests that all states are periodic, in the sense that all his-
tory/future paths are closed. With these assumptions, a probability-based viewpoint
is thereby restored, at the expense of introducing periodicity to quantum systems.
Whether this is just a mathematical trick, or instead has real philosophical implica-
tions, remains an area for future investigation.

7 Conclusions

We have shown that a Berry’s phase-like quantity, Ŵ, can be defined for any quantum
system which undergoes a transition. This “product of amplitudes” is phase invariant,
in the sense that it does not matter what phase choices are made for individual kets.
Each Ŵ is shown to be a quasi-probability associated with a particular closed loop in
time. With this definition, we have shown how the probability of a quantum transition
can be calculated as a sum of Ŵ’s. Quantum interference terms (the so-called “mixed”
terms) are then seen to correspond to those closed loops in which the system does
not self-retrace. Thus, quantum interference can be interpreted as a quasi-probability
effect, and not as traditional interference at all.

Acknowledgements I wish to thank the anonymous reviewers for very helpful comments and sugges-
tions. I also wish to thank William Hodge, Scott Huffman, and William Kerr for interesting discussions.

References

1. Feynman, R.P., Leighton, R.B., Sands, M.: The Feynman Lectures on Physics, vol. III. Addison-
Wesley, Menlo Park (1965)
2. Einstein, A., Podolsky, B., Rosen, N.: Can quantum-mechanical description of physical reality be
considered complete? Phys. Rev. 47, 777 (1935)
3. Bell, J.S.: On the Einstein Podolsky Rosen paradox. Physics 1, 195 (1964)
4. Kirkpatrick, K.A.: “Quantal” behavior in classical probability. Found. Phys. Lett. 16, 199 (2003)
5. Mermin, N.D.: The Ithaca interpretation of quantum mechanics. Pramana 51(5), 549 (1998)
6. Berry, M.V.: Quantal phase factors accompanying adiabatic changes. Proc. R. Soc. Lond. A 392, 45
(1984)
7. Resta, R.: Manifestations of Berry’s phase in molecules and condensed matter. J. Phys.: Condens.
Matter 12, R107 (2000)
8. Loss, D., Schoeller, H., Goldbart, P.: Weak-localization effects and conductance fluctuations: impli-
cations of inhomogeneous magnetic fields. Phys. Rev. B 48, 15218 (1993)
Found Phys (2008) 38: 1073–1081 1081

9. Lyanda-Geller, Y.: Topological transitions in Berry’s phase interference effects. Phys. Rev. Lett. 71,
657 (1993)
10. Zhang, Y., Tan, Y.-W., Stormer, H.L., Kim, P.: Experimental observation of the quantum Hall effect
and Berry’s phase in graphene. Nature 438, 201 (2005)
11. King-Smith, R.D., Vanderbilt, D.: Theory of polarization of crystalline solids. Phys. Rev. B 47, 1651
(1993)
12. Niu, Q., Kleinman, L.: Spin-wave dynamics in real crystals. Phys. Rev. Lett. 80, 2205 (1998)
13. Astumian, R.D.: Adiabatic operation of a molecular machine. Proc. Natl. Acad. Sci. 104, 19715
(2007)
14. Anastopoulos, C., Savvidou, N.: Quantum mechanical histories and the Berry phase. Int. J. Theor.
Phys. 41, 529 (2002)
15. Griffiths, R.B.: Consistent histories and the interpretation of quantum-mechanics. J. Stat. Phys. 36,
219 (1984)
16. Rozanov, Y.A.: Probability Theory. Dover, New York (1969)
17. Bland, J.M., Altman, D.G.: Bayesians and frequentists. Br. Med. J. 317, 1151 (1998)
18. Weizsäcker, C.F. v.: The Structure of Physics. Springer, Dordrecht (2006)
19. Dirac, P.A.M.: The physical interpretation of quantum mechanics. Proc. R. Soc. Lond. 180, 1 (1942)
20. Knight, P.: Quantum mechanics: where the weirdness comes from. Nature 395, 12 (1998)
21. Dennett, D.C.: Intuition pumps. In: Brockman, J. (ed.) The Third Culture: Beyond the Scientific Rev-
olution. Simon & Schuster, New York (1995)
22. Feynman, R.P.: Space-time approach to non-relativistic quantum mechanics. Rev. Mod. Phys. 20, 367
(1948)
23. Everett, H. III: In: DeWitt, B.S., Graham, N. (eds.) The Many-Worlds Interpretation of Quantum
Mechanics. Princeton Univ. Press, Princeton (1986)
24. Tegmark, M.: The interpretation of quantum mechanics: many worlds or many words? Fortsch. Phys.
46, 855 (1998)
25. Zurek, W.H.: Decoherence, einselection, and the quantum origins of the classical. Rev. Mod. Phys.
75, 715 (2003)
26. Raju, C.K.: Time travel and the reality of spontaneity. Found. Phys. 36, 1099 (2006)
27. Gödel, K.: An example of a new type of cosmological solutions of Einstein’s field equations of grav-
itation. Rev. Mod. Phys. 21, 447 (1949)
28. Gott, J.R. III: Closed timelike curves produced by pairs of moving cosmic strings: exact solutions.
Phys. Rev. Lett. 66, 1126 (1991)
29. Kerr, R.P.: Gravitational field of a spinning mass as an example of algebraically special metrics. Phys.
Rev. Lett. 11, 237 (1963)
30. Carroll, S.M., Chen, J.: Spontaneous inflation and the origin of the arrow of time.
arXiv:hep-th/0410270v1 (2004)