A simple Python script gives some insight - here is what I calculate for the sum of the three frequencies:
There is obviously a strong beat at 1 Hz - this is when all three frequencies are (back) in phase. There are minor peaks in between - of which I would consider 5 Hz the most visible component.
This matches intuition - you would expect the difference frequencies of 2 Hz, 3 Hz and perhaps 5 Hz to show up, but they will be muddled. And sure enough, 5 Hz is there; the two bigger peaks around 0.5 s are indicative of the 2 Hz signal; while the 3 Hz signal cannot be separately seen.
Here is the code I used to plot this graph:
# beat frequencies
import numpy as np
import matplotlib.pyplot as plt
from math import pi
f = [101,103,106];
Ns = 8*1024
t = np.linspace(0,2,Ns);
a = np.zeros((Ns),'double')
for fi in f:
a = a + np.cos(2*pi*fi*t)
plt.title('three frequencies beating')