The symbolic nature of Mathematica programs (they are data) makes it easy to transmit from the controlling master kernel to the parallel subkernels. The following example makes a function definition, and then uses DistributeDefinitions to send it to all the subkernels.

test[x_] := Boole[PrimeQ[x, GaussianIntegers -> True]];

DistributeDefinitions[test];

data = ParallelTable[test[x + y I], {x, -50, 50}, {y, -50, 50}];

ArrayPlot[data]