The wave always goes through both slits and the particle goes through only one. The part of the wave that is not associated with a particle is called the empty wave. The particle is guided by the wave (via the quantum potential/velocity from the wavefunction) toward places where the wave density is large and away from places where the density is small.

In the absence of the detectors and , a particle passing through slit 1 (on the left) falls finally on counter , not on , and vice versa. Particles with identical energies and width never cross the axis of symmetry. If the pulsed particle source is so feeble that the particles have different energies, then, of course, the trajectory can cross the axis of symmetry, which is allowed for an asymmetric wavefunction.

If the detectors are inserted, the experimental data shows for a strong measurement ( or ) that the particle traverses the left arm, registered by detector and hits the counter ; while the particle in the right arm, registered by detector hits the counter , and the trajectories do cross. In the case of strong position measuring, the interference fringes disappear. In a complete description of the measuring process, the detectors must also be described by a wavefunction.

Bell [1] stated clearly (twelve years before the discussion about surrealistic Bohm trajectories [2, 3] began) that the naive classical picture does not hold for an isolated quantum system without measurement. The particle, arriving at a given counter, goes through the wrong slit. Reference [2] gives a slightly different approach, but both agree that the position measurement is the key point for understanding the which-way experiment.

In the causal interpretation (CI), there is a nonlocal interaction between the detectors and the whole system, which destroys the superposition of the wavefunctions in the presence of the detectors. For an entangled wavefunction, meaning that the complete overlapping wavefunction cannot be represented as a product state in terms of independent variables, a nonlocal correlation of the particle trajectories appears. In the CI approach the condition for nonlocality is nonfactorizability.

In the Copenhagen interpretation, the measurement collapses the wavefunction. In the CI approach, the position measurement affects the total wavefunction, but in principle it is possible to do a sufficiently subtle path-determining measurement without destroying the interference pattern, which was confirmed by [8] in 2011.

One possible way to solve the problem is to describe the interaction between the measurement device and the isolated quantum system with the conditional wavefunction (cwf) [7, 9]. Here the cwf is the system under observation, a superposition of two entangled wavefunctions, in complete agreement with the Schrödinger equation. One of the wavefunctions defines the measurement device, which here depends only on the time-independent variable (the pointer position), and the other wavefunction defines the isolated quantum system.

In the CI approach, the position-measuring process decomposes the superposition state into a single state, which corresponds to the naive perception of reality, via a nonlocal interaction between the measuring devices triggered by the cwf. So every position registration of a particle, which obeys the time-dependent Schrödinger equation and which is in a superposition of states, is a nonlocal process. In the measuring process, the position of the particle in the direction influences the amplitudes of the detectors in the direction. As a result, measuring the particle's position reduces or cancels out the amplitude of the empty part of the wavefunction.

The graphic shows the possible trajectory, the velocity vector field (red), and the wave density.