When probing for command injection opportunities on a remote host, an attacker might target her own address space from the remote host to verify whether the attack is successful without alerting the target. This helps in verifying a blind attack where the attacker can't see the results directly on the target system, and instead checks if the commands executed on the remote host are affecting her own systems as expected.