Statistical　graph　is　widely　used　to　represent　quantitative　data　in　science,　business,　and　many　other　fields　in　our　daily　life.　To　the　best　of　our　knowledge,　the　neural　mechanisms　underlying　memory　retrieval　based　on　statistical　graphs　have　not　been　reported　yet.　We　investigated　this　issue　using　functional　magnetic　resonance　imaging　(fMRI).　In　our　study,　36　normal　subjects　were　instructed　to　read　statistical　information　that　was　visually　presented　in　either　statistical　graph　or　text.　Then　the　reading　stimuli　disappeared,　subjects　were　instructed　to　answer　a　question　that　needed　to　retrieval　information　of　the　content　of　the　statistical　graph　or　text.　The　direct　comparison　of　the　memory　retrieval　based　on　statistical　graphs　and　that　based　on　text　showed　that　there　were　significantly　more　activated　the　lateral　regions　including　ventral　and　dorsal　pathways,　as　well　as　the　parietal-temporal-occipital　(PTO)　association　area　(BA19/39);　the　medial　regions　including　parahippocampal　cortex　(PHC)　and　retrosplenial　complex　(RSC).　The　results　suggest　that　the　two　visual　pathways　and　the　PTO　are　involved　in　forming　the　imagery　of　statistical　graphs,　and　the　PHC　and　RSC　are　involved　in　the　memory　retrieval　that　played　on　the　mental　imagery　of　statistical　graphs,　in　which　PHC　is　involved　in　the　retrieval　of　object-location　associations　and　RSC　is　involved　in　orientation　mechanisms.　Previous　studies　demonstrated　that　spatial　navigation　was　also　composed　of　two　similar　processes:　retrieval　of　object-location　associations　and　orientation.　We　propose　that　memory　retrieval　based　on　statistical　graphs　draws　on　similar　processing　in　spatial　navigation　that　is　based　on　the　cognitive　maps.　©　2012　IEEE.