"a continuous scale of 1-5" ?

Is it really continuous (all real numbers in [1, 5] are possible) or discrete (values are in {1, 2, 3, 4, 5}) ?

If you want to analyze the interaction of two continuous predictors, it is adviseable to center the predictors. This reduces colinearity, makes the results more stable and leads to interpretable main effects.

To show the results graphically, I would plot the logOdds vs. X1 at some selected values for X2 (and vice versa). I would use the values mean, mean+SD and mean-SD, only when the SD is somehow sensible. Otherwise I would use values that cover the range (min, mean, max, possibly 2 further values).

Another option is to show a contour plot or a heatmap of the logOdds (vesus X1 and X2) .This gives you the complete overview.

You can also show a 3D plot of the logOdds vs. X1 and X2, but the prespective can be tricky. I would prefer the controur plot or heatmap.

Thanks Jochen

That it was indeed helpful 

1) It's a likert 1-5 scale and I have already centered the variables and then analyzed interaction. I have carried out the logodds plot exactly as you said.

2) My question is why not include X2 component while plotting interaction?  I understand that it would just be a constant value for every value of X given how we keep it same for all values of X (either -1SD or +1SD for example). But point being its still a part of the equation B1x1+B2x2+B3x1x2 so shouldn't it be included? Am getting confused because it's not significant. So if I include it will be just how sample behaves and relatable to the population.