Insider attacks are able to evade traditional security controls because the perpetrators of the attack often have legitimate access to protected systems and data. Massive logging of user online activity data (e.g. file access or transfer, use of data storage devices, email records) is collected and analyzed to detect insider attacks (e.g. data theft, fraud, policy violation, etc.). Such techniques are fraught with drawbacks and limitations: 1) the proverbial “needle in a haystack problem,” where very little useful information is found in massive data sets, especially where the incidence of malicious insider activities is very small compared to that of legitimate actors; 2) employee privacy issues may exist about the company monitoring employee behavior; and 3) these techniques are largely wanting in their accuracy, leading to notably high false positive rates. Perhaps the most salient limitation of these techniques is that the analyses are post-hoc, and by the time the activity is detected, the insider has already engaged in data theft or exfiltration, the impact of which may not be reversible. This paper discusses the concept of using probes for detection of threats, wherein user intentions to engage in insider attacks can be gauged by sending carefully designed probes that rouse malicious users into acting. In this research, we seek a broad understanding of the scope and relevance of such probes. There are various motivations for users to steal data, including financial gain, patriotic fervor, and disgruntlement with work. In the present experiment, we created simulated conditions to reflect common insider motivations by providing subjects with imagined scenarios, then asking them to take the perspective of insiders in those scenarios, and explicate their actions through a series of structured questions that mimic our probes. The results show the effect of different scenarios in motivating the users, and the effectiveness of different probes in eliciting their actions.