I've noticed that if I pan something gradually across the whole stereo field, the image doesn't move smoothly between the speakers by my perception. In fact it seems to jump between maybe 7 positions in total. LCR, plus a couple of positions in between each side. Am I normal? 
 
I have always questioned people's decision to pan things by tiny amounts. Like spreading drums out between 5% either side. Does that really make a difference in where those drums are perceived in the stereo image? I can't imagine that it does. Or that it improves separation by any noticeable degree. So what's the point? Surely if you're going to have something that's panned "basically center" then it makes sense to put it slap bang in the middle where the image is going to be the strongest. 

Sounds like your speakers are too far apart. I can pan a mono source from far left to right and I can hear it move smoothly from one side to in the middle and over to the other.
 
There is a technique that is very good for setting speaker distances very well.  It involves a large plank of wood and two friends. You play a mono mix and you start by setting up the speakers on the plank. Then you get the two friends to move them slowly apart while you close your eyes and you listen for the phantom image in the centre. It will be very strong until a point and it will then disappear. Then you just slide them back so the phantom centre image comes back nice and strong again. Then measure the distance remove the plank and reset.
 
Speakers too close is not good either, you get an overstrong centre image and you tend to turn things down that are panned centre. Speakers too far apart tend to make you push centre panned sources up a little too high.
 
Most people are more concerned with computer screens and things in between instead of getting the distance right. Speaker distance first then fit monitors in and around etc..
 
If you don't have this distance right no amount of panning (or pan laws) will ever work.
 
 

pan laws.
 
 
check em
 
set em

It's not your imagination, sharke. The illusion's just not that good to convey small movements accurately. In the real world, we can do it surprisingly well, but that's different. That's sound coming from multiple sources and reflections and being sensed binaurally, not sound coming from just two discrete point sources.
 
Furthermore, we are rarely listening to speakers in the precise center where the illusion is best recreated. More likely, we're listening at a distance, off to one side, or we're using headphones/earbuds that do away with an connection to physical reality altogether.
 
Despite all that, small changes are still useful in mixing. For example, panning harmony vocals 5% off center gives a sense of width that you're missing when they're all panned center. We can't tell they're 5% versus 8%, but we can sense that they're wider than a single point source. Assuming the listener isn't that typical teenager with the one 95-cent earbud. But then, I mix for the 1% - if one person hears it correctly, that's the one person I mixed it for.

I'll admit it is hard to hear a change of 3-4 db in any direction once you start moving off center, especially if you are in a heavy mix, but I do spend quite a bit of time placing things properly in the image.  ("properly" works for us classical guys because the perception of the stereo image is usually generated by a stereo main pair, and the spot mics should fit in that image.) 
 
The panning should be smooth and even unless you have a latency problem and the changes are not kicking in immediately - I sometimes have that problem when inserting a good linear phase eq, it can make any change in the session take a half a second or so to click.  Checking your pan laws can help to make sure the perception of volume remains constant during any change.  

Jeff Evans
Sounds like your speakers are too far apart. I can pan a mono source from far left to right and I can hear it move smoothly from one side to in the middle and over to the other.
 
There is a technique that is very good for setting speaker distances very well.  It involves a large plank of wood and two friends. You play a mono mix and you start by setting up the speakers on the plank. Then you get the two friends to move them slowly apart while you close your eyes and you listen for the phantom image in the centre. It will be very strong until a point and it will then disappear. Then you just slide them back so the phantom centre image comes back nice and strong again. Then measure the distance remove the plank and reset.
 
Speakers too close is not good either, you get an overstrong centre image and you tend to turn things down that are panned centre. Speakers too far apart tend to make you push centre panned sources up a little too high.
 
Most people are more concerned with computer screens and things in between instead of getting the distance right. Speaker distance first then fit monitors in and around etc..
 
If you don't have this distance right no amount of panning (or pan laws) will ever work.
 
 

 
I'm actually quite fastidious about maintaining that equilateral triange between me and the speakers. To this end I have a speaker either side of my main monitor, with my second monitor to the right of the right speaker. It's a bit annoying having to turn my head to see the other monitor but I think it's worth it in terms of stereo imaging. I don't like when the speakers are too wide.  
 

batsbrew
pan laws.
 
 
check em
 
set em

-3dB center, sin/cos taper as always. 

bitflipper
It's not your imagination, sharke. The illusion's just not that good to convey small movements accurately. In the real world, we can do it surprisingly well, but that's different. That's sound coming from multiple sources and reflections and being sensed binaurally, not sound coming from just two discrete point sources.
 
Furthermore, we are rarely listening to speakers in the precise center where the illusion is best recreated. More likely, we're listening at a distance, off to one side, or we're using headphones/earbuds that do away with an connection to physical reality altogether.
 
Despite all that, small changes are still useful in mixing. For example, panning harmony vocals 5% off center gives a sense of width that you're missing when they're all panned center. We can't tell they're 5% versus 8%, but we can sense that they're wider than a single point source. Assuming the listener isn't that typical teenager with the one 95-cent earbud. But then, I mix for the 1% - if one person hears it correctly, that's the one person I mixed it for.

 
I guess maybe I'm just not very good at hearing that subtle width. If I pan things about 15% either way I can start to hear a significant width, but anything less than that....
 
Any mono elements in my mix, I pan them hard left, right or center. The stereo elements I get a little more creative with, like sometimes I'll have two stereo instruments in a mix so I'll pan one from 100L-50L and the other 100R-50R. It sounds less extreme than hard panning in mono. 
 
My perceived width of stereo elements isn't linear either. To my ears, having a stereo instrument panned 50L/50R doesn't sound half as wide as panning it 100L/100R. I would guess it sounds about 70% of the width.

Interestingly enough I just came across this:
 
http://www.moultonlabs.com/more/principles_of_multitrack_mixing_the_phantom_image/P1/
 

Then, a little later still, I got into some loudspeaker research, and found myself called upon one day to make a research recording, wherein I recorded a batch of clicks with very carefully documented changes in level between the stereo channels. This was one of those cases where I figured I knew what was going to happen before I started. Given my golden ears, there just wasn’t much doubt that I could hear the image move as soon as I tweaked the pan-pot even a little, so I decided to calibrate the changes to 1/10th of a decibel, so that I’d be able to really pick out the subtle differences in localization that were going to happen when the levels between channels changed. However, I was very startled to discover that the phantom image didn’t seem to move at all even when the levels between channels changed a whole decibel! I was so startled that I became positive I had made a mistake when preparing the tape! A little investigation (well, about three hours, including chasing down all the wiring in the monitoring system!) showed me that I hadn’t made a mistake, and when the dust finally settled I had found out something quite interesting: that as long as the difference between channels is less than 3 decibels, the phantom image hovers pretty much in the middle point between the two speakers. I promptly ran this down to my buddies at the local loudspeaker factory and we tried it in the anechoic chamber with blindfolds and people pointing at the imaginary phantom, and it still remained true: with up to 3 dB difference between channels (that’s half-power, remember!) the image didn’t move much, maybe five degrees. With between 3 and 6 decibels difference in levels, the phantom quickly and without much stability migrated to the louder speaker, hovering just inboard of that speaker, and once the difference was greater than 7 decibels, the phantom was for all intents and purposes coming from the louder speaker.

Goes to show that you can't be subtle when panning. Hence the popular LCR technique. Personally, I find just 3 positions too limiting but 5 positions is about right: 100L, 50L, C, 50R, 100R. At least for rock and pop. For orchestration, I like 7 positions. What you get then isn't so much a sense of discrete locations but more a general sense of width.