A study that used the field setting you describe is done by Bateson et al. (2006).
As for the mechanism, they write:
we believe that images of eyes motivate cooperative behaviour because they induce a perception in participants of being watched. Although participants were not actually observed in either of our experimental conditions, the human perceptual system contains neurons that respond selectively to stimuli involving faces and eyes (Emery 2000; Haxby et al. 2000), and it is therefore possible that the images exerted an automatic and unconscious effect on the participants’ perception that they were being watched. Our results therefore support the hypothesis that reputational concerns may be extremely powerful
in motivating cooperative behaviour.
Similar findings have been found in a lab setting by Haley and Fessler (2005). They also compared the effect to other cues (such as auditory cues) of presence of others and found similar effect. So I don't have a direct answer to the surveillance camera scenario, but these results suggest the effect would be similar.
However, it is clear that this effect depends on many variables. For example, it has been shown not exist in darkness (Tane and Takezawa, 2011), and to depend on culture (Raihani and Bshary, 2012).
References
- Bateson, M., Nettle, D., & Roberts, G. (2006). Cues of being watched enhance cooperation in a real-world setting. Biology letters, 2(3), 412-414.
- Haley, K. J., & Fessler, D. M. (2005). Nobody's watching?: Subtle cues affect generosity in an anonymous economic game. Evolution and Human behavior, 26(3), 245-256.
- Tane, K., & Takezawa, M. (2011). Perception of human face does not induce cooperation in darkness. Letters on Evolutionary Behavioral Science, 2(2), 24-27.
- Raihani, N. J., & Bshary, R. (2012). A positive effect of flowers rather than eye images in a large-scale, cross-cultural dictator game. Proceedings of the Royal Society B: Biological Sciences, 279(1742), 3556-3564.
