I can think of two ways to do it, neither obvious.
1.
a) use 3dZcutup 185 times to create single slice data
b) use 3dMean to average in groups of 5
c) put back together with 3dZcat
d) use 3drefit to correct the slice width and anything else buggered up in the head file
2.
a) use 3dcalc with differential subscripts to create a dataset where each slice=sum of itself and 4 neighbors. Note that you will still have 185 slices, just spatially summed
b) extract every 5th slice using 3dZcutup
c) and d) as in #1

Here's a little csh script action for #2 that may actually work:
3dcalc -a tom+orig -b a[0,0,-1,0] -c a[0,0,-2,o] -d a[0,0,1,0] -e a[0,0,2,0] -dsZERO -expr "a+b+c+d+e" -prefix avg5
foreach a (`count -d 3 2 185 5`)
3dZcutup -keep $a $a -prefix slice$a
end
3dZcat -prefix newtom slice*HEAD
rm -f slice* avg5+orig*

Tom