ARP cache is the cache which is used by system to temporarily hold IP address to physical address binding, once physical address is found by ARP broadcast then further communication may be done by using it from cache. And this binding is kept in cache for certain amount of time, so ARP cache saves communication time and cost.

When ARP query is broadcasted on network and when reply is sent from destination, then not only transmitter (who originated this query) when all nodes on network find this reply and update their binding table. So any one who has this cache would definitely know IP address to physical address binding of almost all systems. He can simply write a utility/software to just capture all the frames passed on the LAN and this cache will further help him that from which node/session this frame belongs to.

One practical example can be, that any hacker can send ARP response of any network device (may be router or access point) ,so all stations will update their bindings ,and hence in future all the stations will send their packets to that device that may be router or access point. So all the traffic will be captured.